 I used to be a corporate hall, but I'm okay now is the usual introduction I do to public sector folk. I'm very much a private sector refugee, spent 10 years at a massive consumer goods company, which I won't name, but as we said, spent a year in GDS recently and very much enjoyed it. Let's start with the story. Eight years ago I took over as tech lead of this awful financial byddwch chi'n falch y pryd apdysgu y cyflaeth rhannu'r ystyried bunol a hwnnw wedi fel refle dementia. HDMI-yw iawn, mae'r plant gennym ni wedi ddiogelio'i cyfrifiadau mewn Thor. Os efallai, mae'r cyfrifiadau yn allan pen genywb i'i gweithio'r ein fryd yn y cyffraestor, ac mae'n llwyddo i ni'n ddiogelio'r hynny i'r cai iawn a llwyddo i'n cael rwynebu. Mae'r cyfrifiadau sy'n ddirectorio'i cyfrifiadau yn maen nhw, ac mae'r ysgolion, ychydig i gweithio'r ystyried, bydd y dywed, o'r ymgyrch, mae'r cyfnod yn ei wneud i'r gwahanol. Ac mae hefyd yn cael ei byddio'r ysgolion, ac mae'r cyfnod fel ddechrau'n gwahanol. Mae'r meddwl sy'n gweld, mae'n gweld yn gweld ar 11. Mae'n gweithio'n gwahanol oedd yn cyfrifio'r cyfrifio. Mae'n gofyn ymweld sy'n cyfrifio'r Ysgolion ac yn mynd i'r pethau cyfan ychwanegwyr yn cael statiwyr syniad, eu ffordd i dechreu i'r syniad hon i'n gweld flasgwrd yn y systemau ystgril. Rwyf i mi, dwi'n mynd i ffres ydwodi ac mae nhw'n bywys i'r pethau ffarswyr yn y syniad yma. Rydyn ni wneud bod dyna ni'n gwybod themu a chyfthenwyr i'r pethau. Teimlo i'r門hygoel yma. Ond amser wrth yn mynd i ddod, rydyn ni'n ffresywodau. Felly'r sefydlu dyna cofnoddiad, mynd rydyn ni'n megwineb'r ysgrifennu sy'n ddim ddim yng ngynhau, ac efallai rydyn ni'n rhan o'n ddangosol? Felly dyna'r hoffio'r Maid Gymig, mae'n gwahyddwyr wedi bod ni'n weithio'r bywydd, mae'n gwahau'r hwysbryd, mae'r sysadmins eu chwyn hwyro dibynnu'n mhag hwnnw, gyda'r torgu dyna'n ddyliadig. Felly, roedd yn gweithredu, mae'n cydwyno'n dweithio, Myelio gyda'r game. We started having proper testing, testing development. On the opposite side, we learnt to trust a bit better. After the first time we put something live and didn't destroy the world, we got them to agree that we could at least put things out monthly, that's 12 times a year, not just 4 times a year. I suppose the bigger thing was just talking a bit more, which I think we all know that when we try and solve technical problems, usually there are people problems. ac yn amlwg i'w bwysig, a dyna chi'n gweithio'r cyflogwch ar 400 yrhygoffau a rydych chi'n iddynt gael bod yn gweithio'r ysgolwyddiadau ac mae hynny'n iddyn nhw'n gwasanaethu'n gweithio'r cyflogwch ar ffwrdd hynny i gael gwnaeth yr amddangosion, ac mae gennym gweithio'r cyflogwch arall o'r cyflogwch. i'r llwyaf i'r cyfnodd y gofynion, ac fyddwch yn gweithio'r ddau angen i'r ddeithasol a dwi'n gweld i'r llwyffaeth o'r cyfnodd. Fy angen i chi'n bwysig i ddau'r ysgol yn gweld gwnaeth gwaith o'r cyfnodd, ond tewch yn ymgyrch yn ddiddordeb ar y llywodr, a'r ymwysig i'r ysgol yn ymddangos a'r gweithio, ac mae'n ddiddorol yn eich ddau i ddim yn ystod i ddim yn llwyffaeth, a ydych chi'n gweithio i ddim yn ei wneud. Felly, o'r pethau o'r rhan o'r lluniau, ydych chi'n ei ddim yn ei wneud o'r cyfrifysgoriaeth, ond, roeddwn i'r cyfrifysgoriaeth, wedi bod yn gwneud hynny y bydd y rhan o'r sianlitaeth, roeddwn i'n gweithio'r holl yn y cynnig o'r peth, mae'n ddysgu'r cyfrifysgoriaeth i'r sianlitaeth. A'r Deffops mae'n ei ddysgu'r felosifiaeth i'r cyfrifysgoriaeth a'r cyfrifysgoriaeth between those two specialties. It isn't saying that you don't need one or the other or that you can make them the same thing because my God, I was a developer and you did not want me running your production servers. So I think that's really important. It's not about making them the same thing, it's about just improving the collaboration and communication between them. So first of all let's look at some stuff that I can tell you right now won't work if you look at doing this in your organisations. And this is a credit where it's due a really brilliant list from a site called Dev Ops guys. And they've got some really good articles about Dev Ops and what it means for the industry and where things are going, so it's worth looking at. So traps to avoid, don't feel like this is something you can put in a flowchart and inflict on people, that won't work. Agile and Dev Ops aren't the same thing, they work really well together. So if you get your agile development processes and project management going, then the Dev Ops philosophy and way of working can be a really wonderful complement to that, but don't confuse the two things. Re-branding a team, or either a Dev team or Ops team and just calling them a different name, isn't going to work. And starting a separate Dev Ops group isn't going to work either, because really what this is about is a philosophy of reducing silos rather than creating new ones. And I think the last one's the thing I see the most often, which is Ops teams going, well it's all about Ops, so we'll own this now and Dev teams trying to do the same thing. And that's probably, many of you are leaders in your organisations, that's probably the one I'd say, watch out the most of all, don't let this become a power struggle, it needs to be about finding a new way of working together. Believing it's a buzzword is another thing that really doesn't work or that it's a silver bullet, it's really not going to solve all your problems. As I said earlier, it's really not about Dev's managing production or Dev driven release management. But I do think that everybody can make this happen. It doesn't, you don't need special people, you don't need special time to do it in, you don't have to wait for a crisis and probably waiting for a crisis is the worst thing to do. But, you know, there's a lot of times when you talk to organisations about why best practice would be good for them and their immediate answer is just, but we're different, we can't do it that way. That was a bit depressing, so let me talk about what does work. This is a framework that John Willis has put out, which I think works very well and anybody who has read the GDS service design manual, this is a framework you'll see there as well, which is to start with culture but also look at automation, measurements and sharing. So, what do we mean by culture? It's all about people and reducing that us and them mentality that Dev's broke production and it's their fault, but focusing on fast and stable. And that's really the most important thing about Dev Ops as a philosophy is it's a way of smoothing the road so that your business that needs faster and faster change, which we've all the other speakers have spoken about a lot today, and you want to be able to do things more frequently, add new functionality, respond to the users, to your citizens, your internal folks, but you also have a public out there that expects things to be up not just 99.9, but 99.9x9% of the time. And so this is a way of getting to be agile and fast in your change, but also stable in the service that you provide. Invest in automating everything that you can. There's very few, as admins I know, who really want to sit every day and stroke tin, as we were talking about earlier, or their job be about monitoring alerts or seeing what's broken and then fixing it. In giving time to invest in making, working on the system to reduce time working in the system is really, really valuable. And you look at some of the most interesting, to me, startups these days, they can spin up new servers, services, and so much in clicks of a button, or even automatically. And that's something which, with my big systems outsourced private sector background is just a dream. I couldn't believe how well that was being done when I joined GDS, and you literally saw a board up on the wall that ran all the tests for every part of Gov UK in under three minutes. And any change that was deployed, they could see whether it had broken anything in three minutes. And being from a world where a full range of testing and a project took three weeks, I was extremely impressed with that. The other thing is to measure things. We all talk about you get what you measure, but it's even more true when you look at operations, knowing what you break when you break it, being able to respond well to things like anonymous targeting you, DDoS tax and similar. But again, it's an investment to be able to put some of that measurement, the analytics and so on in place, so that you can then focus on improving the things that are needed. And the last is sharing, this is a terribly saccharant Haley Joll Osmond movie, but the cultures that we're seeing that are really adopting DevOps as a philosophy very well tend to be really good at sharing back, not just inside their own companies, but with the broader community. There's a series of meetups called DevOps days that started in Belgium and have now spread around, and so if the people in your organisation who are looking after your development and your operations aren't hooked into that kind of community and seeing what's going on there, then it's possibly worth investing in getting them some access to some of the thinking that's going on here. As I said earlier, GDS is pretty impressive in this regard that that philosophy of developers and operations folks working closely together. The approach of continuous delivery, having plenty of monitoring and automation. My good friend, Gareth Rushgrove, has a series of t-shirts all about how much he loves graphs and more and more people buying them for him as pranks now. I mean that I thought this from James Thornett, who's the product manager for GovUK, that in the first six months of GovUK, there were over a thousand releases. That's the sort of thing that is almost incomprehensible if you're used to that more traditional way of doing operations and doing development. The fact that a thousand code changes had been released in that six month period is pretty impressive. My other favourite thing about GDS is they have a GDS badger of deployment that tweets when they do releases, which I think tells you a little bit about the DevOps culture at GDS as well. In terms of a bit further reading about this, I've posted all of these on my blog as well, so there's a load of different posts about anti-patterns and patterns. A guy called Nick Bottolomeos has written a really amazing presentation about how you introduce this culture in a very traditional legacy bound environment. And so probably that's the one area that I didn't feel I had time to talk through in detail today, but I'd highly recommend that presentation to those of you who are finding yourself with loads of existing transactional systems, whether you own them or you're working with organisations on them that need to be brought into the 21st century in terms of the way that development and operations is done. As mentioned by a few people, the GDS's digital service manual is excellent and there's certainly a good page there about how to approach this as well. And there's a weekly newsletter, which I'd also recommend that you have a look at. Given the time, it was really just a bit of a taster of what this is and hopefully some food for thought in terms of how it might be useful for you guys, but hopefully it was good. Thank you very much.