Inside Star Citizen: Meshing Together

Transcript — every row opens the video at that moment

1. 00:00:00

   [Music] meetings hands down the most exciting element in all of game development now everyone that has ever made a game will tell you that it's gathering in a meeting room and talking about stuff that's really their very favorite part of the process yes the best days in Game Dev are the ones full of meetings and the best behind the scenes content is always incredibly meeting focused now imagine a summit an entire week dedicated to hashing out the plans for the next month quarter year and sometimes Beyond yes exciting stuff I know all kidding aside Summits are an essential element of Game Dev we don't often get to explore so when we heard

2. 00:00:48

   that the core technology Network and engine teams were coming together to discuss how best to move from persistent entity streaming into tackling the immense server meshing we figured you might just indulge Us in trying something a little different this week so we set up some cameras and grabbed a few willing participants just minutes after completing their second day of summits to get a candid unfiltered look inside one of the most important Star Citizen Summits of 2023 and let me tell you January has been full of them so to spare you the full effect of five days of meetings we'll be spreading these out over the next few weeks while continuing our look at content being developed for 319 and Beyond so after just a little Ado we present to

3. 00:01:37

   you the server meshing Strike Team quarter one 2023 development Summit post meeting review name pending do you need me to look into the camera off the site right okay okay um yeah yeah there you go I didn't know what you know I can't I can't do this if they're watching me and go away what are you guys doing here uh well I'm visiting the uks who discuss

4. 00:02:27

   server meshing with the rest of the team we're doing a summit to try and plan for the next deliveries of pyro and Silver Machine so Summits in general especially this one is really where all the teams connect because you know with the pandemic a lot of people have been working from home so some has become really important because it allows us to reconnect together so we meet in one location uh this time it's in the UK and we have a representative of every team so most of the leads and every team it's the first time that we're meeting as a team for a lot of us a lot of people joined the company a lot of people join those teams and they've never met so far so it's just great like we have someone from Korea Montreal have people from all over you know so it's also a strong

5. 00:03:16

   aspect of the senates this is a great opportunity for us this is the first time ever we've had the Strike Team all in one place all at one time this is our kind of Avengers assembled moment for Server meshing this is the time we get three core teams together to actually work out um the final nuances of the architecture to build this system there are many teams are involved and to get the teams on the same page yeah we're just building an emblems in a room talk for one week aligning all gods that we can just go out and crank it to court without less communication than we would have without them many people already working together for a long time so it's communication easier if you know each other and getting forward to yeah equating on goals based on that we can more efficiently execute two or two ago because especially in a global company where time zones uh it's kind of hard

6. 00:04:05

   sometimes I think about software developments like oil tanker where it once put into One Direction but as soon as you want to change direction it takes a long time because you need to get everyone has a slightly different idea of how stuff works and to get everyone really understand how stuff should work on a more detailed level there's a lot of work and align it once everybody's put their stuff in we plan it out we discuss approaches strategies on how we're going to intertwine our work and this allows us to be on the same page so that we can then work together to achieve a very specific objective and not just be building technology for fun the objective of the Summit is to flesh

7. 00:04:55

   out the exact road map for Server meshing based on all of the work we've been doing before PES so PS is the state of the simulation that's now persisting that's great behind us now we need to actually server mesh the major goal of this Summit is to reflect back on what we've done and see how we can do it better for the next Milestones on server machine what did we forget what can we do better how can we better organize ourselves so we keep moving forward and we don't do what we've had to do a lot for the previous delivery which is think and create as you go and there will always be a little bit of this in

8. 00:05:42

   game development it's normal you need to have this creativity you need to keep that but we also need to get a little bit better at planning and anticipating some of the challenges the goal of the me of the Summit is really to flesh out the steps to get there with the objective to ship a second solar system that's really the primary objective what do you call a jump point for you because we are talking a lot of the low technical yeah that's what I want to know yeah a secondary goal is talking about goodbye secondary goal would be talking about the low technical groundwork for to enable features like jump point so this is day two I think of the summit

9. 00:06:36

   so we've we've done a first retrospective of all of the PES work issues we've encountered problems we've had things we want to avoid and this so this is officially the first day of actually digging into missions I know the second thing um day two uh we have drunk so much coffee day two has been super great in that we've we've aligned a lot of our different roadmaps every you know every team gets to have their own sort of view on the project and I think today we really came together as a this is the critical path that we need to achieve everything else is gravy but we have the critical path is really clearly identified which is a big deal in the size of this project surf meshing the plan for persistent anti-streaming

10. 00:07:23

    and surf mesh they were designed to be done simultaneously at the same time this is going back to late 2020 um we couldn't do that it just wasn't practical in the end so we had to devote all our attention to doing persistent entity streaming um this was in the club like during lockdown so we did all of this while we're all working from home uh we're now in a position uh with the the surf meshing stuff um and I kind of lost my train of thought I was going to say something really good enough for growing worse yeah we know what the critical bet is which is the that's the most important thing in these technology projects because they get to be very large and so sometimes you lose track of what the actual executive is and our objective is to get a second solar system online as fast as possible not just to build server meshing technology but it's to get the second solar system online now

11. 00:08:10

    we have a rough road map of the different steps we need to do to get our second solar system online and so now we start arguing about the later steps and the strategies after that we're tackling a lot of really difficult problems at the moment server meshing is one of the toughest problems to solve in video to game development and uh we've gone to the point now we're getting to persistent entity streaming uh to the point where it's shippable uh that then kind of opens up lots of opportunities to be able to get this this quite impressive system built so yeah this uh there's good stuff Happening Here so the first the first phase is the separation out of the replication layer that unlocks the Pyro jump gate stuff that also unlocks the surf meshing stuff uh we've been talking about the uh kind

12. 00:08:59

    of uh zones and Authority but you're much better off talking to a proper engineer about this kind of stuff uh but yeah we're we're getting a good handle on this stuff we're getting a good plan for delivery technology and now we need to build on what we built there defining more sub steps to go out to actually bring power on the hands of the player this is a really complicated system and we had to learn how to do it we there's been a lot of experimentation there's been a lot of invention when we had our initial plan at the beginning of uh 2021 and late 2020. there was a lot of stuff in there that we knew we had to build but didn't quite know how to build it when we actually came to actually build those interesting systems we had to kind of re-found problems we found stuff that was never expected we found limitations

13. 00:09:48

    to the technology we found limitations of what the hardware could do so we very much had to get our Engineers to invent our way around the problem so this is very much a feedback this is a learning experience to figure out how we need to build surf meshes the key thing that we're doing with this is serve meshing is also tied into the power the delivery the jump gates to get to pyro um it's not quite the same problem but it also kind of is when you actually do a jump to get to pyro where you have to transfer Authority from the Stanton dgs to the Pyro dgs um what we want to do of course is have multiple dgs persist star system but we're separating the actual jump gate stuff from the actual uh surf meshing stuff so that we get to the point so we can actually assure pyros delivery and whilst we worry about the

14. 00:10:35

    actual uh server meshing delivery so we're actually yeah those are now from our perspective two different problems so very early on uh we're going to be tackling uh what what we're calling separation of the replication layer out from the dgs so that's the thing which replicates all our data it allows communication it's a single source of Truth in memory for this day of the entire universe at any given time that's going to get separated out from the dgs that currently holds all the fun simulation stuff that's happening at the moment uh that will that's a first key step both to delivering pyro all the technical back end for the jump Gates and it's also a kind of first big step for Server meshing uh yeah so that's what we're going to be tackling initially the design that we're putting in now

15. 00:11:24

    we've had for a couple of years and we've been refining it over that time and the time we're building building persistent entity streaming there have been bits and pieces of surf meshing on there that we needed to get to get done uh but we're now in a position now to go really uh now we've got system entity streaming down there as a kind of foundational piece that then opens up the opportunity to really kind of uh get this design that we've had that we've found um to really buckle down and actually get it uh dashed out and get the first version out a static server machine like this this is a piece of work that's going to last for quite a long time there'll be ongoing stuff for quite a while moving to the goalie even five six years ago when Mr object container streaming where we started to split up the game into smaller chunks and then moving from there having stuff only partial load on the clients having stuff partially load on the server and then decoupling the

16. 00:12:13

    persistent State and the next steps are through this tablet more into more logical units to execute the code which actually server matching this though gets really a continuation and over time you the more you understand that's the problem you're more you know what you need to do so um yeah that's why every year kind of most likely we have a summit once per year on this topic for the last 10 years to refine what we do and yeah what exactly we need to do after we go to step Neo that's the critical path to the first step but after that is is how we get to the actual end game which is the dynamic mesh so we all have our views on this but now it's really we're getting out into the grave view like how is this gonna work here's cases we have to deal

17. 00:13:00

    with like what if a player dies into a jump Tunnel right what happens to the missions right uh what happens to the persistent state of these tunnels do they exist do they not exist do they respond so we talk about necessarily the gameplay aspect of it like I'm I'm giving you guys example now but really the technology aspect of it so do these zones keep existing how do we reuse them do we want to reuse them uh what's the effect on PES right for for this Zone transition as subsystems so that's been the highlight of it has been really about identifying that and then how we make it dynamic the best thing which came out of today is the same which comes again out of

18. 00:13:49

    summits it's to get alignment that everyone gets the same idea into their mind in a detailed level this is after day two I think the biggest question is still how we gonna how we're gonna operationalize the dynamic mesh and so uh you know we still have these big statements like oh we're going to request additional dgs's and then assign them now we're gonna have to flesh out how do we request them how do we know they're ready how do we assign them how does that actually work uh is going to be the key to to the next steps that will then be built over the coming months and of course it's not just us as well as the Strike Team we have uh what 50 or so game teams that are also going to take this code and then have to adapt their all their functionality that they

19. 00:14:37

    have across uh Star Citizen to use all this cool technology so yeah this is the start of server meshing in this curve form in our case it's cramming a lot of developers into a room uh closing the doors getting to drink coffee and talking a lot it's very much a list of stuff that we've got to do determine we basically have a list of stuff that we need to nail down open questions miscommunications between teams we need to get these three key teams all aligned all pointed in the same direction so they're all fully understand where it is they're building we resolve the technical niggles that we have between these teams so they really understand what it is they're making we correct problems before they happen so we there's misunderstandings within the interfaces between the teams that be quite expensive to fix having a summit like this allows us to fix these problems in advance and fix them much

20. 00:15:25

    much easily so yeah we need to just uh this is what gives us the solid plan that we're going to be working to for the next year so touchwood we're in a position now to finally push on and get the very first version which is static server mesh in place we're about to find other problems on this journey but we can invent OA we know we can invent our way around these problems so uh yeah we may occasionally get slowed down but we don't get uh we're still determined to deliver this so what did we learn this week well we learned that PES is not just the technical Harbinger we all knew it would be it was also the critical experience necessary to review amend and evolve existing plans for server meshing with the Lessons Learned along the way that it has a variety of Technologies including the long-awaited jump points

21. 00:16:12

    that will enable the addition of pyro and star systems Beyond and that people really boy do they really love to just sit down for an interview just five minutes after sitting in intense all-day meetings and let me tell you I owe some people some dinners now we'll be back with more traditional ISC features and return visits to the meshing Strike Team Summit in the coming weeks and don't forget to keep your eyes on the robertspace industries.com website and socials for information on the pinning release of alpha 318. it's got Salvage cargo persistence and more and it seems there's a new Scorpius the things you'll learn on Twitter for inside Star Citizen I'm Jared Huckabee and we'll see you all here next week