[00:00.000 --> 00:09.000]  Thanks very much. I leave you in the capable hands of Karen Jax.
[00:09.000 --> 00:22.000]  Hi, thanks, Jimmy. Hi, so I'm Karen Jax and I'm going to talk to you today about DBA evolution
[00:22.000 --> 00:27.000]  or the changing role of the database administrator. So you can see that my title at the moment
[00:27.000 --> 00:32.000]  isn't database administrator, but I was a DBA for over 20 years.
[00:32.000 --> 00:36.000]  I'm not going to talk through the agenda. I've just left it there so that you can keep track of where we're up to
[00:36.000 --> 00:40.000]  and when I might stop talking.
[00:40.000 --> 00:45.000]  So I promise that this would be a talk for everyone that's pondered life's important questions.
[00:45.000 --> 00:50.000]  What does a DBA actually do? Why are DBAs always so grumpy?
[00:50.000 --> 00:54.000]  How's the role of the DBA changed over the last couple of decades?
[00:54.000 --> 01:01.000]  What's the DBA of the future going to do? And will autonomous databases finally have put us all out of work?
[01:01.000 --> 01:07.000]  So this year marks the 25th anniversary of the start of my career as a DBA,
[01:07.000 --> 01:13.000]  so I thought that I'd allow myself a bit of a self-indulgent retrospective.
[01:13.000 --> 01:18.000]  So first, the evolution of this particular DBA, and I'm going to apologize to Claire at this point
[01:18.000 --> 01:22.000]  because this is going to be a bit of a me, me, me section of my talk.
[01:22.000 --> 01:26.000]  So I'm just going to have to get through that bit.
[01:26.000 --> 01:29.000]  Being a DBA definitely wasn't an ambition of mine at school.
[01:29.000 --> 01:35.000]  I mean, anybody here say database administrator when they were asked what they wanted to do when they grew up?
[01:35.000 --> 01:42.000]  I mean, I didn't even know I was going to go into IT until about a year before I started my career,
[01:42.000 --> 01:46.000]  although with hindsight it's fairly obvious that that's what I was destined to do.
[01:46.000 --> 01:51.000]  My dad taught me to count in binary before I could even properly count in decimal.
[01:51.000 --> 01:57.000]  He built me my first computer, an acorn atom, which he later replaced with a BBC Model B,
[01:57.000 --> 02:02.000]  and obviously I used the computer for playing Chuckie Egg and other games,
[02:02.000 --> 02:06.000]  but I also learned to code in BBC Basic.
[02:06.000 --> 02:12.000]  And I actually looked up the spec of the BBC Model B in case anyone's interested,
[02:12.000 --> 02:19.000]  so it had a 2MHz processor and 64K of RAM.
[02:19.000 --> 02:25.000]  I then went on to do a maths degree because I still had absolutely no idea what I wanted to do,
[02:25.000 --> 02:27.000]  and maths was my favourite subject.
[02:27.000 --> 02:30.000]  And after that, I figured I probably ought to get some real-world skills,
[02:30.000 --> 02:33.000]  so I did a masters in software development.
[02:33.000 --> 02:38.000]  Towards the end of that, one of the graduate schemes I applied to did some aptitude tests,
[02:38.000 --> 02:43.000]  and they came back to say that they thought I'd probably be good as a DBA.
[02:43.000 --> 02:49.000]  Once I'd asked Jeeves what one of those was, I thought that sounded like a fairly good fit,
[02:49.000 --> 02:52.000]  and that I'd give it a go.
[02:52.000 --> 02:57.000]  Fortunately, it turns out that Jeeves was right, and the aptitude test,
[02:57.000 --> 03:01.000]  and this is what my careers looked like so far,
[03:01.000 --> 03:06.000]  so I really haven't strayed very far from my DBA role.
[03:06.000 --> 03:11.000]  Obviously, the things that I do in my job have changed over time
[03:11.000 --> 03:14.000]  and I've gradually moved into more senior roles,
[03:14.000 --> 03:17.000]  and I've gradually changed the focus of what I do,
[03:17.000 --> 03:20.000]  but I've tried not to let that influence what I talk about too much.
[03:20.000 --> 03:25.000]  I'm trying to talk more about the DBA role in general.
[03:25.000 --> 03:31.000]  So, let's set the scene for the start of my career as a DBA in 1998.
[03:31.000 --> 03:34.000]  The 90s brought us Grunge and Britpop.
[03:34.000 --> 03:39.000]  They also brought us the first truly portable and affordable mobile phones,
[03:39.000 --> 03:42.000]  for example, the Nokia 3210.
[03:42.000 --> 03:45.000]  They brought us the first text message, the Palm Pilot,
[03:45.000 --> 03:51.000]  and also for better or worse, Amazon and Google.
[03:51.000 --> 03:55.000]  But what does a DBA actually do?
[03:55.000 --> 03:57.000]  Just so that we're all on the same page,
[03:57.000 --> 04:02.000]  I've taken highlights of definitions of a DBA from various places.
[04:02.000 --> 04:07.000]  I looked at careers services, DBMs vendors, and job sites.
[04:07.000 --> 04:13.000]  And the general consensus is that a DBA manages and secures computer systems
[04:13.000 --> 04:17.000]  that store data using specialist software.
[04:17.000 --> 04:19.000]  So, they're all pretty much agreed,
[04:19.000 --> 04:26.000]  but that doesn't really tell us anything about what a DBA actually does.
[04:26.000 --> 04:32.000]  So, first of all, one thing to note is that there are lots of different types of DBA.
[04:32.000 --> 04:36.000]  And the different types of DBA will be different from organisation to organisation.
[04:36.000 --> 04:46.000]  So, at ACME, for example, we might have the DBA role split into many different DBA roles.
[04:46.000 --> 04:53.000]  Whereas at Wonka Industries, we might have DBAs who are in one pool
[04:53.000 --> 04:58.000]  who are all responsible for all DBA tasks.
[04:58.000 --> 05:04.000]  A traditional split is between the production DBA and the development DBA.
[05:04.000 --> 05:08.000]  So, the production DBA will be responsible for keeping all of the databases
[05:08.000 --> 05:10.000]  in the production environment up and running.
[05:10.000 --> 05:14.000]  So, looking at things like availability, performance, security.
[05:14.000 --> 05:18.000]  The development DBA would work more closely with the developers.
[05:18.000 --> 05:22.000]  So, they've got a focus on building and maintaining the database environment
[05:22.000 --> 05:29.000]  that's supporting the application development lifecycle.
[05:29.000 --> 05:31.000]  Just wait for the photos.
[05:31.000 --> 05:37.000]  In some organisations, you might see a split of system versus application DBA.
[05:37.000 --> 05:45.000]  So, here the application DBA will be more looking at the logical application related aspects of the database.
[05:45.000 --> 05:50.000]  Whereas the system DBA is going to be responsible for the underlying software
[05:50.000 --> 05:54.000]  and the physical infrastructure.
[05:54.000 --> 05:57.000]  And the splits not clear cut.
[05:57.000 --> 06:00.000]  The roles often overlap. They may or they may not.
[06:00.000 --> 06:04.000]  And you might see lots of different types of DBA as well.
[06:04.000 --> 06:11.000]  So, you might have a data warehouse DBA who will look after the data warehouses
[06:11.000 --> 06:16.000]  particularly ETL from various data sources into the organisation's data warehouse.
[06:16.000 --> 06:20.000]  A cloud DBA I think I've got in there as well
[06:20.000 --> 06:27.000]  who will be looking after cloud hosted databases including liaising with the cloud provider.
[06:27.000 --> 06:30.000]  Also, you might see database architects.
[06:30.000 --> 06:36.000]  I know it doesn't have DBA in the title but it's a very closely related role and often overlaps.
[06:36.000 --> 06:40.000]  It's a role that's existed for at least a couple of decades
[06:40.000 --> 06:49.000]  but it's gradually become more popular and seems in a lot of cases to have split out from the traditional DBA role.
[06:49.000 --> 06:54.000]  And it's not an exhaustive list. You could see all sorts of different types of DBA.
[06:54.000 --> 07:01.000]  You might also get DBAs that look after just one particular task such as backup and recovery or replication.
[07:01.000 --> 07:07.000]  And any or all of those might be split into junior DBA and senior DBA
[07:07.000 --> 07:11.000]  with the junior DBA doing a lot of the day-to-day tasks
[07:11.000 --> 07:15.000]  and the senior DBA having more of a strategic focus.
[07:15.000 --> 07:25.000]  Okay, so we saw the definition of a DBA role
[07:25.000 --> 07:29.000]  but as I said it didn't really tell us much about what a DBA actually does.
[07:29.000 --> 07:32.000]  So what does a DBA do in 2023?
[07:34.000 --> 07:38.000]  I looked at the different definitions that I showed you earlier
[07:38.000 --> 07:41.000]  and there were lists of responsibilities that went with those
[07:41.000 --> 07:44.000]  and I've taken information from different job adverts
[07:44.000 --> 07:48.000]  and tried to pull together the list.
[07:48.000 --> 07:53.000]  So most people seem to agree that a DBA will do some or all of
[07:53.000 --> 07:59.000]  designing, implementing and managing backup and recovery policies
[08:01.000 --> 08:06.000]  designing and implementing security policies and managing database access
[08:06.000 --> 08:11.000]  although in a lot of places the security role has been split out into a separate security team these days.
[08:11.000 --> 08:16.000]  Implementing monitoring and ongoing monitoring of the databases
[08:16.000 --> 08:22.000]  design and development of the database including data modelling
[08:22.000 --> 08:28.000]  or maybe just reviewing data models that have been created by the development team.
[08:28.000 --> 08:35.000]  Support and troubleshooting including often being on a 24-7 on-call rotor
[08:35.000 --> 08:42.000]  and database software install and upgrades listed in a lot of them as a specific task
[08:42.000 --> 08:49.000]  but in a lot of places that's automated or the installation side of things is automated
[08:49.000 --> 08:53.000]  so the DBA would either be responsible for implementing that automation
[08:53.000 --> 08:56.000]  or another team would do the automation
[08:56.000 --> 08:59.000]  and the DBA wouldn't actually have anything to do there.
[08:59.000 --> 09:02.000]  Upgrades I would say it's more of a strategic role
[09:02.000 --> 09:08.000]  in working with the application team to make sure that upgrades are tested correctly
[09:08.000 --> 09:10.000]  so that they're successful.
[09:10.000 --> 09:16.000]  And there's a whole other page, a whole other list of responsibilities
[09:16.000 --> 09:19.000]  so providing database expertise to other teams
[09:19.000 --> 09:23.000]  performance tuning enhancement often troubleshooting
[09:23.000 --> 09:29.000]  capacity planning creating databases including for development and test environments
[09:29.000 --> 09:34.000]  database maintenance and data protection and GDPR considerations.
[09:36.000 --> 09:39.000]  So that was a massive, oh sorry, before I move on
[09:39.000 --> 09:41.000]  is a massive long list of responsibilities
[09:41.000 --> 09:45.000]  so clearly in most cases one person can't do all of those things.
[09:48.000 --> 09:53.000]  This chart shows some of the most requested skills for a DBA in the UK
[09:53.000 --> 09:58.000]  I've taken this from IT Jobs Watch so it's for the six months up to January 2023
[09:58.000 --> 10:05.000]  so we can see that SQL is the first requested skill
[10:05.000 --> 10:11.000]  and I went up to number 23 because there we've got PostgreSQL listed there as well.
[10:11.000 --> 10:16.000]  On the next page I've summarised those different responsibilities
[10:16.000 --> 10:19.000]  sorry the different skills that are being asked for.
[10:21.000 --> 10:25.000]  So consolidating the list of skills it boils down to
[10:25.000 --> 10:33.000]  SQL so writing SQL, making it more efficient
[10:33.000 --> 10:36.000]  potentially knowing procedural languages
[10:36.000 --> 10:39.000]  integration and everyone's favourite ORMs.
[10:41.000 --> 10:46.000]  One or more database management systems usually still asking for relational.
[10:49.000 --> 10:54.000]  Knowledge of operating systems so this will be traditional operating systems
[10:54.000 --> 10:57.000]  but also various different cloud environments.
[10:57.000 --> 11:02.000]  Performance tuning, database migration seems to be quite a popular thing
[11:02.000 --> 11:05.000]  people are always moving from one database to another.
[11:05.000 --> 11:10.000]  Disaster recovery, social skills is listed most of the places as a separate thing
[11:10.000 --> 11:12.000]  I don't know what they're trying to say there.
[11:16.000 --> 11:20.000]  HA replication clustering type skills
[11:20.000 --> 11:25.000]  and then not listed in any of the things I found as specific skills
[11:25.000 --> 11:30.000]  but most DBAs I know are actually expected to know about different DevOps tools
[11:30.000 --> 11:33.000]  and methodologies and they're expected to know how to use automation tools
[11:33.000 --> 11:35.000]  like Ansible or Puppet to do things.
[11:39.000 --> 11:43.000]  So how does all that compare to what a DBA did in 1998?
[11:44.000 --> 11:46.000]  On paper it actually looks very similar
[11:46.000 --> 11:50.000]  so there are just reproduced the same list of responsibilities as I had
[11:50.000 --> 11:52.000]  for what a DBA does in 2023.
[11:52.000 --> 11:55.000]  Although I've taken off GDPR because I didn't exist yet.
[11:57.000 --> 11:59.000]  Does that mean nothing's changed?
[12:00.000 --> 12:06.000]  Well obviously although the responsibilities are pretty much the same
[12:06.000 --> 12:12.000]  the skills that a DBA needs and the tools that a DBA is expected to be able to use
[12:12.000 --> 12:17.000]  have changed massively because a lot of the tasks that they do day to day
[12:17.000 --> 12:19.000]  are done very differently.
[12:20.000 --> 12:24.000]  And obviously that's not limited to DBA tasks or even IT tasks
[12:24.000 --> 12:27.000]  we do loads of day to day tasks differently as well.
[12:27.000 --> 12:30.000]  So I've used send my sister a birthday card as an example
[12:30.000 --> 12:35.000]  so that's the flowchart for 1998 I would have gone to the shop
[12:35.000 --> 12:38.000]  I would have browsed the racks, chosen a card I liked
[12:38.000 --> 12:41.000]  I would have paid for that card and probably bought a stamp at the same time
[12:41.000 --> 12:43.000]  taking it home, written a message in it
[12:43.000 --> 12:46.000]  put my sister's address on it, stuck the stamp on
[12:46.000 --> 12:53.000]  and then walked it to my local post box to be collected by the post delivery.
[12:53.000 --> 12:57.000]  I can still use that method in 2023 if I want to
[12:57.000 --> 12:59.000]  but this is what I'm more likely to do
[12:59.000 --> 13:03.000]  I'm going to open the app of my favourite online greetings card retailer
[13:03.000 --> 13:08.000]  I'm going to upload a photo from my phone, type in a message on my sister's address
[13:08.000 --> 13:11.000]  pay using my disposable online credit card
[13:11.000 --> 13:15.000]  and just get them to send it directly to my sister as soon as it's printed.
[13:20.000 --> 13:23.000]  So if we go back to DBA tasks
[13:23.000 --> 13:26.000]  I've taken installed database software as an example
[13:26.000 --> 13:29.000]  I've not put this there to be read
[13:29.000 --> 13:34.000]  this is an extract from the Oracle 7.3 installation guide
[13:34.000 --> 13:37.000]  which was my first database installation that I did
[13:37.000 --> 13:41.000]  I can't imagine that many DBAs in 2023
[13:41.000 --> 13:45.000]  need to first walk to the machine room and put the CD into the server
[13:45.000 --> 13:47.000]  before they start their installation
[13:47.000 --> 13:53.000]  it used to take me I think two days to prepare and perform a single database install
[13:53.000 --> 13:56.000]  and we didn't have a repeatable process
[13:56.000 --> 14:00.000]  there was a silent mode that was supposed to allow us to script it
[14:00.000 --> 14:04.000]  but because none of the servers were installed using any kind of automation
[14:04.000 --> 14:09.000]  they had their own particularities so it didn't work and we didn't bother with it
[14:09.000 --> 14:15.000]  in 2023 a DBA would it be expected to know how to use an automation tool
[14:15.000 --> 14:19.000]  such as Ansible to do a database software install
[14:19.000 --> 14:22.000]  and they'd probably not just install a standalone database
[14:22.000 --> 14:26.000]  they can install the whole database infrastructure
[14:26.000 --> 14:31.000]  including high availability monitoring back up and recovery
[14:31.000 --> 14:34.000]  all in one command
[14:39.000 --> 14:43.000]  so what kind of skills did a DBA need in 1998?
[14:45.000 --> 14:48.000]  I can't reproduce the list of skills quite as easily
[14:48.000 --> 14:52.000]  because that has changed but a lot of the general skills were the same
[14:52.000 --> 14:57.000]  so you'd be expected to know SQL and probably a procedural language
[14:57.000 --> 15:04.000]  you'd be expected to know one relational database management system
[15:04.000 --> 15:10.000]  you'd probably be expected to know how to do Unix basic admin and shell scripting
[15:10.000 --> 15:14.000]  database performance tuning and disaster recovery
[15:14.000 --> 15:20.000]  which was probably calling someone to bring the tapes back on site
[15:20.000 --> 15:28.000]  so that's what a DBA was doing in 1998 and is doing today
[15:28.000 --> 15:32.000]  what will a DBA do in another 25 years time?
[15:34.000 --> 15:37.000]  well I'm going to hedge that question for now
[15:37.000 --> 15:42.000]  so before we start looking at what a DBA might do in 25 years time
[15:42.000 --> 15:44.000]  everyone likes a bit of buzzword bingo
[15:44.000 --> 15:48.000]  so I thought we'd have a look at the buzzwords, trends, hypes
[15:48.000 --> 15:52.000]  from the last few decades just to give us an idea of what's changed
[15:52.000 --> 15:56.000]  around our vintage DBA and what's still changing
[15:57.000 --> 16:00.000]  so the 1990s
[16:00.000 --> 16:05.000]  the 90s brought us the worldwide web first with painfully slow dial-up access
[16:05.000 --> 16:10.000]  and finally broadband internet which meant that remote work was finally a possibility
[16:10.000 --> 16:15.000]  for some DBAs even if it was just to avoid having to trek into the office
[16:15.000 --> 16:18.000]  when doing 24-7 on-call
[16:18.000 --> 16:21.000]  the concept of big data was introduced
[16:21.000 --> 16:25.000]  data warehouses became popular and we needed massively parallel processing
[16:25.000 --> 16:30.000]  MPP to analyse all of this data so teradata became popular
[16:30.000 --> 16:33.000]  and a teaser didn't come along until 2000s
[16:33.000 --> 16:37.000]  scrum started to creep into our normal everyday work process
[16:38.000 --> 16:42.000]  object-oriented programming and from their object-oriented databases
[16:42.000 --> 16:46.000]  were the next big thing in programming methodology
[16:46.000 --> 16:49.000]  everybody was talking about AI and machine learning
[16:49.000 --> 16:51.000]  even though they weren't new concepts
[16:51.000 --> 16:55.000]  with deep blue-beating Grandmaster Kasparov at chess
[16:55.000 --> 16:58.000]  we were blessed with PostgreSQL and Linux
[16:58.000 --> 17:02.000]  SSDs started to replace spinning hard drives
[17:02.000 --> 17:04.000]  spinning disk hard drives
[17:04.000 --> 17:08.000]  but that's still a gradual ongoing move
[17:08.000 --> 17:11.000]  and a lot of us missed out on New Year's celebrations
[17:11.000 --> 17:14.000]  because we were on call to make sure that planes didn't fall out of the sky
[17:14.000 --> 17:17.000]  and fortunes weren't wiped out due to the Y2K bug
[17:20.000 --> 17:22.000]  the 2000s
[17:22.000 --> 17:25.000]  we were introduced to blockchain
[17:25.000 --> 17:28.000]  Cloud was and still is one of the biggest buzzwords
[17:28.000 --> 17:31.000]  with a move away from the traditional data centre
[17:31.000 --> 17:34.000]  everything started to be provided as a service
[17:34.000 --> 17:37.000]  we had software as a service and platform as a service
[17:37.000 --> 17:40.000]  we were all expected to be agile to embrace DevOps methodologies
[17:40.000 --> 17:44.000]  and to implement continuous integration and continuous deployment
[17:45.000 --> 17:50.000]  the Internet of Things and social media contributed to a massive growth in data volumes
[17:51.000 --> 17:54.000]  we got Hadoop, Exadata, NoSQL and JSON
[17:54.000 --> 17:58.000]  as new ways to store, process, analyse and describe data
[17:59.000 --> 18:02.000]  BASE was proposed as an alternative to ACID
[18:02.000 --> 18:07.000]  favouring availability over consistency for distributed systems
[18:07.000 --> 18:11.000]  access to data on the move started to be possible
[18:11.000 --> 18:14.000]  with Wi-Fi and containers were introduced
[18:14.000 --> 18:17.000]  although they weren't at this point quite ready for databases
[18:18.000 --> 18:21.000]  I'm going to take a deep breath and drink some water
[18:21.000 --> 18:36.000]  OK, 2010s infrastructure joined the as a service party
[18:36.000 --> 18:39.000]  NFTs, non-fungible tokens were introduced
[18:39.000 --> 18:42.000]  we started to store our data in data lakes
[18:42.000 --> 18:44.000]  which sometimes became data swamps
[18:45.000 --> 18:49.000]  columnar databases, even though they were actually a concept from the 1970s
[18:49.000 --> 18:52.000]  became popular and joined the other NoSQL databases
[18:53.000 --> 18:57.000]  we started to avoid putting all of our eggs in one basket with multi-cloud strategies
[18:58.000 --> 19:00.000]  Oracle announced their autonomous databases
[19:00.000 --> 19:03.000]  which meant there was no, not going to be any need for DBAs anymore
[19:04.000 --> 19:09.000]  CloudNative was the next big thing with microservices, Docker and Kubernetes
[19:09.000 --> 19:13.000]  and the EU caused DBAs all across Europe
[19:13.000 --> 19:19.000]  massive headaches by introducing the General Data Protection Regulations GDPR
[19:22.000 --> 19:26.000]  the 2020s haven't all been a bed of roses in tech so far
[19:26.000 --> 19:29.000]  with a global pandemic and massive layoffs
[19:29.000 --> 19:34.000]  on the other hand, remote work has started to become the norm for a lot of us
[19:34.000 --> 19:39.000]  access to data on the move is getting faster with the introduction of 5G
[19:39.000 --> 19:43.000]  Zoom became a household name
[19:43.000 --> 19:45.000]  everyone's now heard of Master Don
[19:45.000 --> 19:48.000]  and apparently we're all going to be living in the metaverse
[19:48.000 --> 19:52.000]  and using chat GPT to write all of our reports and presentations
[19:53.000 --> 19:55.000]  and that's just the first few years
[19:55.000 --> 19:59.000]  so who knows what's going to have happened by the end of the 2020s
[19:59.000 --> 20:16.000]  so, we can look at a few ways that databases have changed in response to all of those things that have been going on
[20:19.000 --> 20:24.000]  this is a massively simplified view of the history of database management systems
[20:24.000 --> 20:26.000]  please don't quote me on this
[20:26.000 --> 20:32.000]  so in around 1980 the relational database pretty much
[20:32.000 --> 20:36.000]  started to replace the old navigational databases
[20:36.000 --> 20:39.000]  and it's been going strong since then and still is
[20:39.000 --> 20:46.000]  in about 2000, various post-relational or no-SQL databases
[20:48.000 --> 20:52.000]  became popular and are still popular
[20:52.000 --> 20:59.000]  and of course everything goes in cycles, the no-SQL databases were around before the relational databases
[20:59.000 --> 21:02.000]  just that they weren't called no-SQL databases at that point
[21:05.000 --> 21:08.000]  it's easy to think that relational databases are going away
[21:08.000 --> 21:12.000]  when we hear about all the different no-SQL databases there are and how popular they are
[21:12.000 --> 21:18.000]  this is a chart from dbengens.com from January this year
[21:18.000 --> 21:22.000]  and it shows that actually relational databases are more than twice as popular
[21:22.000 --> 21:28.000]  as all of the different types of database management system put together
[21:30.000 --> 21:35.000]  and their figures also show that the popularity of relational database management systems
[21:35.000 --> 21:38.000]  has stayed fairly static over the last 10 years
[21:41.000 --> 21:44.000]  we've got more variety of databases
[21:44.000 --> 21:48.000]  so in the 90s there was a handful of database management systems
[21:48.000 --> 21:54.000]  and organisations would usually use one of those for all of their databases or maybe two
[21:54.000 --> 21:58.000]  there are now over 400 different types of database management system
[21:58.000 --> 22:04.000]  most organisations don't restrict themselves to just one or two of those
[22:04.000 --> 22:09.000]  so you'll get not just within one organisation multiple database types
[22:09.000 --> 22:14.000]  but also within one application
[22:14.000 --> 22:17.000]  so that means a DBA might be expected to know
[22:17.000 --> 22:20.000]  not just several relational database management systems
[22:20.000 --> 22:24.000]  but also various different no-SQL types they might be expected to know
[22:24.000 --> 22:29.000]  about spatial databases, time series, graph databases or document store
[22:32.000 --> 22:35.000]  databases have got bigger
[22:35.000 --> 22:39.000]  so the terms big data and very large database were introduced in the 90s
[22:39.000 --> 22:43.000]  but they don't actually tell us very much about the specific volume of data
[22:43.000 --> 22:48.000]  the definition in most places I found of very large database
[22:48.000 --> 22:52.000]  is a database that stores so much data
[22:52.000 --> 22:59.000]  that its maintenance and architecture need specialist tools and methodologies
[22:59.000 --> 23:06.000]  so it's basically just a database that's so big that we haven't yet figured out how to manage it
[23:06.000 --> 23:12.000]  so in the 90s most people I spoke to seemed to think that a multi gigabyte database was fairly big
[23:12.000 --> 23:17.000]  these days people have multi terabyte databases and they're seen as fairly normal
[23:17.000 --> 23:21.000]  once you start to get multi tens of terabytes you might start needing
[23:21.000 --> 23:29.000]  some specialist methodologies
[23:29.000 --> 23:38.000]  in 2018 IDC said that the global data sphere was predicted to grow to 175 zettabytes by 2025
[23:38.000 --> 23:45.000]  I haven't managed to find any later figures so I don't know if that's on course to still be the case
[23:45.000 --> 23:52.000]  so the global data sphere is apparently all of the data that's created, captured and replicated
[23:52.000 --> 23:55.000]  in core edge and end point locations
[23:55.000 --> 23:59.000]  and one zettabyte is a trillion gigabytes
[23:59.000 --> 24:05.000]  so for anyone like me who has difficulty visualising 175 trillion gigabytes
[24:05.000 --> 24:08.000]  apparently if you burnt that onto CDs
[24:08.000 --> 24:13.000]  you'd have a stack of CDs that either took you to the moon 23 times
[24:13.000 --> 24:17.000]  or around the earth 222 times
[24:17.000 --> 24:26.000]  so it doesn't look likely that there's going to be any shortage of data to manage any time soon
[24:26.000 --> 24:28.000]  databases have got more complex
[24:28.000 --> 24:34.000]  I suspect we've all seen an application architecture diagram that looks approximately like that
[24:34.000 --> 24:39.000]  we've got multiple database types
[24:39.000 --> 24:45.000]  not just within one organisation but also often for a single application
[24:45.000 --> 24:48.000]  we've got multi cloud strategies
[24:48.000 --> 24:50.000]  we've got containerised databases
[24:50.000 --> 24:52.000]  we've got distributed systems
[24:52.000 --> 25:00.000]  and we've got many interrelated components
[25:00.000 --> 25:07.000]  so how has the DBA adapted in response to all of these changes?
[25:07.000 --> 25:11.000]  the DBA has had to develop a cloud mindset
[25:11.000 --> 25:15.000]  I was really reluctant to embrace cloud
[25:15.000 --> 25:18.000]  I can't remember when I first heard the word
[25:18.000 --> 25:22.000]  but I thought it was just a hype that was all going to blow over
[25:22.000 --> 25:26.000]  whether we like it or not cloud is here and it's growing
[25:26.000 --> 25:32.000]  so we have to work with it
[25:32.000 --> 25:39.000]  so with the advent of cloud computing many DBAs are no longer actually responsible for managing on-premise databases
[25:39.000 --> 25:43.000]  the hardware and software is often managed by somebody else
[25:43.000 --> 25:52.000]  so the cloud DBA will need to have knowledge of probably multiple different cloud platforms
[25:52.000 --> 25:58.000]  and even organisations that keep their databases on-premises
[25:58.000 --> 26:07.000]  are often moving to a cloud-like architecture, a cloud-like model with them self-service databases
[26:07.000 --> 26:09.000]  the DBAs had to learn to be flexible
[26:09.000 --> 26:14.000]  so as I've said they would be expected often to know multiple different types of database
[26:14.000 --> 26:17.000]  not just relational but also no SQL
[26:17.000 --> 26:22.000]  they need to be able to work in lots of different database environments
[26:22.000 --> 26:30.000]  not only the traditional operating systems but also cloud and things like Kubernetes
[26:30.000 --> 26:36.000]  they're expected to know how to use DevOps and automation tools and methodologies
[26:36.000 --> 26:44.000]  and they've got to keep up to date with all the changes in all of these different types of tool and methodology
[26:44.000 --> 26:51.000]  the other side of the coin is that actually a lot of roles have been separated out into different teams
[26:51.000 --> 26:58.000]  and you might now have different teams that are responsible for infrastructure, for data, security and the application
[26:58.000 --> 27:01.000]  we already looked at the different types of DBA
[27:01.000 --> 27:07.000]  but actually there are a whole host of different roles related to database and the data in it
[27:07.000 --> 27:12.000]  and this is just a few of the ones that I've seen by scanning job sites
[27:12.000 --> 27:21.000]  all of these changes mean that the DBA has to collaborate with other teams
[27:21.000 --> 27:28.000]  so not just with other database experts but also with teams that support the entire application stack
[27:28.000 --> 27:32.000]  I often say to people that I've been forced out of my little database bubble
[27:35.000 --> 27:39.000]  and the DBAs had to develop a strategic focus
[27:39.000 --> 27:45.000]  so the DBA is more likely to be looking at the overall architecture of the database
[27:45.000 --> 27:54.000]  analytics and data processing, improving automated processes, collaborating with other teams
[27:54.000 --> 27:57.000]  to make sure they've implemented business solutions
[27:57.000 --> 28:01.000]  rather than looking at the day-to-day technical details of managing the database
[28:01.000 --> 28:09.000]  so it's a reminder that change isn't all bad
[28:09.000 --> 28:15.000]  I asked colleagues and contacts to tell me some of the DBA tasks that they used to have to do
[28:15.000 --> 28:19.000]  that they are glad to be rid of and these are just a few of the answers
[28:19.000 --> 28:27.000]  vacuum cron jobs, data file reorganizations, commuting to work
[28:27.000 --> 28:31.000]  controversial ones, some people probably do still have to do that
[28:31.000 --> 28:38.000]  writing complicated backup scripts and installing post-GIS via scripts rather than create extension
[28:38.000 --> 28:43.000]  so you've probably all got your own tasks that you are glad to be rid of
[28:47.000 --> 28:52.000]  so the million dollar question, a variation of the question is the DBA role
[28:52.000 --> 28:58.000]  obviously is asked constantly all over the internet and of course database conferences
[28:58.000 --> 29:03.000]  and probably more so since Oracle announced its autonomous database in 2017
[29:05.000 --> 29:12.000]  is the DBA the clock winder of the future or the lamp lighter, the switchboard operator, the video rental clerk
[29:12.000 --> 29:15.000]  or probably more likely the COBOL developer
[29:15.000 --> 29:20.000]  because even if things change so much that modern databases don't require a DBA
[29:20.000 --> 29:27.000]  how many decades is it going to take before all databases are migrated over to this new DBA-less technology
[29:32.000 --> 29:43.000]  fortunately data from the US Bureau of Labor Statistics says that there isn't a shortage of DBA roles
[29:43.000 --> 29:56.000]  so we can see this goes from 2003 so 2003 is the first year in which those stats listed database administrator as a separate role
[29:56.000 --> 30:03.000]  before that I think it was computer scientists or I can't remember the exact role
[30:03.000 --> 30:06.000]  but there wasn't a specific DBA role
[30:06.000 --> 30:13.000]  from 2020 it's listed as database administrators and architects
[30:13.000 --> 30:17.000]  so that supports what we saw before about the two roles splitting out from the same thing
[30:18.000 --> 30:26.000]  so this is the total number of DBA's thousands across the left there in the US
[30:26.000 --> 30:29.000]  so we can see that it's going up slightly
[30:29.000 --> 30:34.000]  it's not growing massively but it's definitely a slight upward trend
[30:34.000 --> 30:42.000]  and the Bureau actually projects a 9% growth from 2021 to 2031
[30:42.000 --> 30:45.000]  which is higher than the average of all other occupations
[30:46.000 --> 30:47.000]  so that's good
[30:49.000 --> 30:56.000]  I tried to find similar stats for the UK but unfortunately the lists of job roles are more generic
[30:56.000 --> 30:58.000]  they don't have database administrator listed
[30:58.000 --> 31:06.000]  so instead I've gone to IT Jobs Watch and I've found the numbers of job adverts that have DBA
[31:06.000 --> 31:08.000]  that are for DBA or database administrator
[31:08.000 --> 31:12.000]  and those numbers are looking pretty healthy for 2022 and 2023
[31:15.000 --> 31:17.000]  well how about automation?
[31:17.000 --> 31:20.000]  surely once everything's automated we don't need DBA's
[31:21.000 --> 31:24.000]  so one of my friends is a pilot and people like to ask him
[31:24.000 --> 31:26.000]  well the plane just flies itself doesn't it?
[31:26.000 --> 31:28.000]  you just put it in autopilot
[31:28.000 --> 31:30.000]  you don't really need to be there as a pilot
[31:30.000 --> 31:33.000]  and he says well yeah if everything's going well that's fine
[31:33.000 --> 31:38.000]  you could just have someone with a lot less experience sitting there watching the dials
[31:39.000 --> 31:41.000]  but as soon as something goes wrong
[31:41.000 --> 31:43.000]  as soon as there's any kind of emergency situation
[31:43.000 --> 31:45.000]  you want someone there that knows what they're doing
[31:47.000 --> 31:51.000]  okay it's not usually such a life or death situation if a database has an emergency
[31:51.000 --> 31:54.000]  but you want someone on hand that knows what they're doing
[31:54.000 --> 31:56.000]  to fix that and get things up and running again
[31:57.000 --> 32:00.000]  so you still need someone there to make the decisions
[32:02.000 --> 32:06.000]  and actually if each database takes less time to manage
[32:06.000 --> 32:10.000]  that means that one DBA can manage more databases
[32:10.000 --> 32:12.000]  well we've seen how much more data there is
[32:12.000 --> 32:14.000]  and a similar number of DBA's
[32:14.000 --> 32:17.000]  so that seems to probably work well
[32:20.000 --> 32:22.000]  what about the junior DBA's?
[32:22.000 --> 32:24.000]  I've not been able to find any figures
[32:24.000 --> 32:27.000]  but I wonder whether there are fewer DBA's
[32:27.000 --> 32:29.000]  fewer junior DBA roles
[32:29.000 --> 32:32.000]  if automation is replacing a lot of the tasks
[32:32.000 --> 32:35.000]  the day-to-day tasks that junior DBA's used to do
[32:35.000 --> 32:37.000]  do we still have roles for them?
[32:37.000 --> 32:39.000]  and if that's the case
[32:39.000 --> 32:42.000]  how can somebody become a senior DBA
[32:42.000 --> 32:44.000]  if they haven't been a junior DBA
[32:44.000 --> 32:46.000]  does that mean that the career path is changing?
[32:46.000 --> 32:50.000]  and back to autonomous databases
[32:50.000 --> 32:52.000]  I mean if we believe the headlines
[32:52.000 --> 32:57.000]  when Oracle announced autonomous databases back in 2017
[32:57.000 --> 32:59.000]  DBA's were soon going to be a thing of the past
[32:59.000 --> 33:01.000]  there was absolutely no need for them
[33:01.000 --> 33:07.000]  so even if all databases were moved over to autonomous databases
[33:07.000 --> 33:12.000]  we've got the day-to-day maintenance and patching is automated
[33:12.000 --> 33:17.000]  but you still need somebody to design the databases
[33:17.000 --> 33:19.000]  to do the data modelling
[33:19.000 --> 33:22.000]  to make sure that there's efficient code accessing the database
[33:22.000 --> 33:26.000]  and actually if you look at some of the reviews
[33:26.000 --> 33:28.000]  these are just extracts from a couple of recent ones
[33:28.000 --> 33:32.000]  apparently it's an extremely complicated thing to put in place
[33:32.000 --> 33:35.000]  so you still need somebody who's an expert
[33:35.000 --> 33:38.000]  to be able to put that infrastructure in place in the first place
[33:38.000 --> 33:44.000]  so fortunately the answer to all of that
[33:44.000 --> 33:47.000]  was no the DBA role isn't obsolete
[33:50.000 --> 33:53.000]  so what conclusions can we draw from all of this?
[33:53.000 --> 33:57.000]  the DBA role has changed since 1998
[33:57.000 --> 33:59.000]  and it will continue to change
[33:59.000 --> 34:04.000]  in response to changes in database technology and methodologies
[34:04.000 --> 34:11.000]  the DBA of 2023 has similar responsibilities to the DBA of 1998
[34:11.000 --> 34:13.000]  but does them in a very different way
[34:13.000 --> 34:16.000]  and so they're expected to be able to do very different things
[34:16.000 --> 34:20.000]  relational databases are still going strong
[34:20.000 --> 34:24.000]  they're still the majority and we're not expecting them to go anywhere
[34:24.000 --> 34:28.000]  but DBAs will be expected to know about other databases as well
[34:30.000 --> 34:33.000]  the amount of data is growing exponentially
[34:33.000 --> 34:35.000]  being managed by a similar number of DBAs
[34:35.000 --> 34:39.000]  so that suggests that we really do still need those DBAs
[34:41.000 --> 34:45.000]  and things are moving either to the cloud
[34:45.000 --> 34:50.000]  or to a cloud like infrastructure on premises
[34:54.000 --> 34:57.000]  so the main thing for me is don't panic as we've seen
[34:57.000 --> 34:59.000]  quantity of data is growing
[34:59.000 --> 35:02.000]  a lot of that data is going to be stored in databases
[35:02.000 --> 35:05.000]  and that's going to have to be managed somehow
[35:05.000 --> 35:07.000]  even if things are automated
[35:07.000 --> 35:09.000]  you want someone on hand in an emergency
[35:09.000 --> 35:12.000]  you want somebody to design those databases
[35:12.000 --> 35:16.000]  you want somebody to write and improve the automation code
[35:16.000 --> 35:20.000]  you want someone to fix the interesting queries
[35:20.000 --> 35:23.000]  that have been created by ORMs
[35:23.000 --> 35:25.000]  so it's a role that's evolving
[35:25.000 --> 35:27.000]  like any other role in IT
[35:27.000 --> 35:29.000]  like any other role anywhere else
[35:29.000 --> 35:31.000]  but it's not going away
[35:31.000 --> 35:33.000]  if you're happy to embrace automation
[35:33.000 --> 35:35.000]  and to try new things
[35:35.000 --> 35:38.000]  you've got a long and happy career in front of you as a DBA
[35:39.000 --> 35:42.000]  and even if you're a stick in the mud old-fashioned DBA
[35:42.000 --> 35:44.000]  and you don't want to change
[35:44.000 --> 35:46.000]  well there are going to be enough legacy systems
[35:46.000 --> 35:48.000]  to keep you in work for a long time to come
[35:48.000 --> 35:50.000]  so you'll be fine until you retire
[35:50.000 --> 36:00.000]  that's it
[36:04.000 --> 36:06.000]  I got through that faster than I expected
[36:06.000 --> 36:09.000]  so I've got time for questions
[36:09.000 --> 36:19.000]  so just a note from my side
[36:19.000 --> 36:22.000]  what I discovered is very interesting for me
[36:22.000 --> 36:24.000]  is the operator concept in Kubernetes
[36:24.000 --> 36:28.000]  and instead of using a managed PostgreSQL database
[36:28.000 --> 36:29.000]  which is expensive
[36:29.000 --> 36:32.000]  I just use an operator and the operator itself
[36:32.000 --> 36:36.000]  it doesn't just install a database automatically for me
[36:36.000 --> 36:39.000]  it also does piggy-bounds
[36:39.000 --> 36:41.000]  and it does switching from left to right
[36:41.000 --> 36:43.000]  and I think this will be the future
[36:43.000 --> 36:45.000]  so a lot of knowledge we have right now
[36:45.000 --> 36:47.000]  will be moved into an operator
[36:47.000 --> 36:50.000]  it's more central and it will be much better for us
[36:50.000 --> 36:51.000]  I agree
[36:51.000 --> 36:54.000]  if you're going to have a containerized environment
[36:54.000 --> 36:57.000]  operators are definitely the way to go
[36:57.000 --> 36:59.000]  the problem that I've seen
[36:59.000 --> 37:01.000]  and I guess because of the role I do
[37:01.000 --> 37:03.000]  I see when there are problems
[37:03.000 --> 37:04.000]  is that people think
[37:04.000 --> 37:06.000]  that does everything, we don't need DBAs
[37:06.000 --> 37:10.000]  so you still need that DBA knowledge behind it
[37:10.000 --> 37:12.000]  definitely
[37:12.000 --> 37:15.000]  so working in smaller companies
[37:15.000 --> 37:18.000]  the DBA is basically if you're lucky
[37:18.000 --> 37:21.000]  the DevOps team and if you're unlucky
[37:21.000 --> 37:24.000]  the developer team which means
[37:24.000 --> 37:26.000]  a lot of people don't have the knowledge
[37:26.000 --> 37:28.000]  or the interest or time
[37:28.000 --> 37:32.000]  and things are done in a sort of not proper way
[37:32.000 --> 37:35.000]  so as somebody who is interested
[37:35.000 --> 37:37.000]  but has a day job as a developer
[37:37.000 --> 37:40.000]  how can you foster that environment
[37:40.000 --> 37:43.000]  where developers will start to understand
[37:43.000 --> 37:47.000]  if they're making inefficient queries
[37:47.000 --> 37:49.000]  or if they're building the database wrong
[37:49.000 --> 37:52.000]  and get them to both know and be interested
[37:52.000 --> 37:56.000]  and also get the company to kind of care about those things
[37:56.000 --> 37:59.000]  rather than keeping them until it's an emergency
[37:59.000 --> 38:02.000]  like you said then the plane is crashing down
[38:02.000 --> 38:04.000]  That's a really good question, yeah
[38:04.000 --> 38:07.000]  so I wish I had all of the answers to that
[38:07.000 --> 38:09.000]  because I've been in lots of organizations
[38:09.000 --> 38:12.000]  as a DBA or helping
[38:12.000 --> 38:16.000]  and we've had exactly that situation
[38:16.000 --> 38:19.000]  so I would say one thing is make friends
[38:19.000 --> 38:21.000]  make sure that all teams
[38:21.000 --> 38:24.000]  are actually talking to each other
[38:24.000 --> 38:28.000]  as a DBA I was always kind of trying to
[38:28.000 --> 38:31.000]  integrate the other teams and actually be useful
[38:31.000 --> 38:34.000]  I mean obviously sometimes I was that typical grumpy DBA
[38:34.000 --> 38:37.000]  no you can't do that, no you can't have all these permissions
[38:37.000 --> 38:39.000]  I was enforcing rules
[38:39.000 --> 38:42.000]  but I tried really hard to be there
[38:42.000 --> 38:45.000]  and helpful and show that I could actually
[38:45.000 --> 38:47.000]  help to make this query run faster
[38:47.000 --> 38:49.000]  or help to make that better
[38:49.000 --> 38:51.000]  so I think that's...
[38:51.000 --> 38:54.000]  and I did training courses for our developers as well
[38:54.000 --> 38:57.000]  to teach them about the basics of databases
[38:57.000 --> 39:00.000]  that's great because I found that most developers
[39:00.000 --> 39:03.000]  are really open to that because they want to learn
[39:03.000 --> 39:08.000]  as far as people higher up in the organization go
[39:08.000 --> 39:11.000]  I'm not sure, I don't know what the answer is there
[39:11.000 --> 39:14.000]  but probably just showing them how it benefits the organization
[39:14.000 --> 39:16.000]  is the only way to go there
[39:19.000 --> 39:21.000]  Hello, thank you for your talk
[39:21.000 --> 39:26.000]  I feel completely, I agree completely with your talk
[39:26.000 --> 39:28.000]  I'm a DBA as well
[39:28.000 --> 39:30.000]  I have a couple of questions
[39:30.000 --> 39:34.000]  First thing is the role of the DBA
[39:34.000 --> 39:37.000]  is still an old male role
[39:37.000 --> 39:41.000]  so what can we do for making it more attractive
[39:41.000 --> 39:43.000]  for the woman?
[39:43.000 --> 39:46.000]  because it's something very stressful
[39:46.000 --> 39:50.000]  it's something that is very stressful
[39:50.000 --> 39:51.000]  another thing
[39:51.000 --> 39:53.000]  another million dollar question
[39:53.000 --> 39:56.000]  it's something that is not changed yet
[39:56.000 --> 39:57.000]  it's still old male
[39:57.000 --> 40:01.000]  and I don't think it's a good idea to continue this way
[40:01.000 --> 40:03.000]  I know it's an horrible job
[40:03.000 --> 40:06.000]  but how can we make it more attractive
[40:06.000 --> 40:08.000]  and the other thing is
[40:08.000 --> 40:11.000]  a lot of companies, a lot of my customers
[40:11.000 --> 40:13.000]  struggle to find DBA
[40:13.000 --> 40:16.000]  how can they make it more attractive
[40:16.000 --> 40:19.000]  for getting a permanent DBA
[40:19.000 --> 40:22.000]  because I give advice of hiring DBA
[40:22.000 --> 40:24.000]  because it's important to having in-house
[40:24.000 --> 40:26.000]  but they struggle to find DBA
[40:26.000 --> 40:28.000]  thank you
[40:28.000 --> 40:31.000]  The first question I wish I knew the answer to that
[40:31.000 --> 40:36.000]  I have no idea how to get more women into being DBAs
[40:36.000 --> 40:42.000]  that's something that's a problem across the whole of the IT community
[40:42.000 --> 40:45.000]  and something that's not changed since I started 25 years ago
[40:45.000 --> 40:47.000]  the only things that I do
[40:47.000 --> 40:53.000]  is I go into part of an organisation that does coding workshops for girls
[40:53.000 --> 40:55.000]  to try and get people in early
[40:55.000 --> 40:59.000]  I mean I was so lucky because my dad was a self-taught coder
[40:59.000 --> 41:01.000]  he loved IT
[41:01.000 --> 41:04.000]  he built me my first computer and showed me how to play with it
[41:04.000 --> 41:06.000]  that's what people need, they need to see that it's fun
[41:06.000 --> 41:08.000]  I mean, well, we think it's fun
[41:08.000 --> 41:10.000]  I know it's not everybody's idea of fun
[41:10.000 --> 41:13.000]  I know, my husband still doesn't understand
[41:13.000 --> 41:16.000]  and I say no, I know but it's my hobby as well
[41:16.000 --> 41:18.000]  so I think that's it
[41:18.000 --> 41:21.000]  and that's probably the same as that first question
[41:21.000 --> 41:24.000]  is that we need to make this look like fun to people
[41:24.000 --> 41:28.000]  and maybe we need to try and get rid of that image of the one PDBA, I don't know
[41:28.000 --> 41:31.000]  We need to clean up our ugly aspect
[41:31.000 --> 41:34.000]  Exactly, more flashing badges
[41:40.000 --> 41:42.000]  Hello, thank you very much for the talk
[41:42.000 --> 41:45.000]  I wanted to ask you, I'm a data engineer myself
[41:45.000 --> 41:47.000]  I'm a data engineer myself
[41:47.000 --> 41:52.000]  and I feel that it's complicated to get across what we do every day
[41:52.000 --> 41:56.000]  because we don't do any very visual things
[41:56.000 --> 42:01.000]  we're not web developers so we can show really nice pages or nice graphics
[42:01.000 --> 42:07.000]  Do you have any advice to give for somebody who's in the field of databases
[42:07.000 --> 42:13.000]  and behind the scenes to make their work more noticeable?
[42:13.000 --> 42:16.000]  Yeah, that's a very good question
[42:18.000 --> 42:22.000]  I don't know, I suppose it depends which audience you're looking at
[42:22.000 --> 42:25.000]  are you looking at internally within an organisation
[42:25.000 --> 42:30.000]  or outside to the general public?
[42:30.000 --> 42:34.000]  internally but for people who don't know much about it
[42:34.000 --> 42:37.000]  people who do other things that manage it
[42:37.000 --> 42:43.000]  Yeah, I think some of it is showing the value that they bring
[42:43.000 --> 42:47.000]  the problem with a lot of things, I can speak for being a DBA
[42:47.000 --> 42:50.000]  is that when everything's going right you're not noticed
[42:50.000 --> 42:53.000]  because what's the DBA doing?
[42:53.000 --> 42:55.000]  We're not with nothing's going on
[42:55.000 --> 42:59.000]  we all need to start putting little things in the code
[42:59.000 --> 43:02.000]  to cause problems every so often
[43:02.000 --> 43:08.000]  so maybe just try and bring attention to the value that you bring
[43:08.000 --> 43:12.000]  Claire is probably a very good person to talk to later
[43:12.000 --> 43:15.000]  because I don't know if you saw Claire's talk earlier
[43:15.000 --> 43:18.000]  about how to write things in blogs
[43:18.000 --> 43:22.000]  so that you get your point across clearly and that kind of thing
[43:22.000 --> 43:32.000]  Claire has some very good ideas on how to communicate