Making an independend MMO - The Albion Online Story
Download as PPTX, PDF
2 likes2,165 views
Talk at devcom game developer conference in Cologne, August 2018. Discusses the design, development and operation of the MMORPG Albion Online.
Making an independend MMO - The Albion Online Story
- 1. MAKING AN INDEPENDENT MMO THE ALBION ONLINE STORY DAVID SALZ (DAVID@SANDBOX-INTERACTIVE.COM) ROBIN HENKYS (ROBIN@SANDBOX-INTERACTIVE.COM)
- 2. WHO ARE WE? David Salz, CTO and co-founder of Sandbox Interactive 18 years in game development Founder of Indie-Studio Bitfield & BitEngine (only commercial engine for Nintendo DS) Robin Henkys, CEO & Game Director 11 years in game development Previously worked on Drakensang Party RPGs and was the lead designer on Drakensang Online, a hack & slay MMO
- 3. WHO IS SANDBOX INTERACTIVE? An independent game developer founded in 2012 Based in Berlin, Germany Employs 30 full time on-site plus another 15-20 part time Freelancers worldwide Exists to develop, market and distribute Albion Online
- 4. WHAT IS ALBION ONLINE? Independent MMORPG privately funded and self-published! > 500.000 players Open Beta in 2016, Release in 2017, Steam-Release in 2018 Sandbox, Full-Loot, PvP and Guild-heavy One World (like Eve Online) Cross-platform (PC/Mac/Linux/Android/iOS)
- 13. FOUNDING THE COMPANY External investor with game idea Good budget, but way too small for the suggested scope Won the pitch! Suggested inexpensive prototype Modesty and honesty pays off! good: used an existing indie team as a starting point of the team already existing network was invaluable!
- 14. GETTING STARTED core team had never done an online game... (but lots of experience otherwise) got external coaches with relevant experience! picked proven technology / middleware (a first for all of us!) made some hard choices in game design to avoid production risks simple art style (small team, small budget, performance considerations) loading screens instead of streaming world
- 18. MIDDLEWARE Unity3D (starting w/ 3.5 in 2012, trying to always use the latest stable) Photon (Cross-Platform Networking library and server framework) using C# on client and server (with lots of shared code) Cassandra high-performance NoSQL in-memory database offers horizontal scaling! Postgres best (feature-wise) free SQL DB
- 19. KEEPING CONTROL OF THE TECHNOLOGY good: used only minimal feature set of engine, network middleware kept Unity completely out of server! (way too slow, way to unreliable!) fully functional "command line" client that works without Unity (also allows bots, easy load tests etc.) wrote key components ourselves: level loading, collision, pathfinding, AI, complete server structure etc. using Unity for Graphics, Animation, Visual Effects, UI, Input handling good compromise between using an engine and roll-our-own!
- 23. PROTOTYPE & REFINANCE Initial Prototype proved ability of the team Led to additional investment & continued dev into Alpha This was a change to the original plan
- 24. COMMUNITY BUILDING involved large guilds in the development process at a very early stage tested early builds with closed user groups in 2-4 week "alpha" test sessions offered special rewards such as custom guild logos etc. invaluable feedback became the seed of our community with long reaching loyalty
- 26. KEEPING DESIGN ON TRACK Took radical design decisions based on the idea that these would have emotional resonance Set these design decisions as immoveable pillars and stuck with them: All items player crafted Full Loot PvP You are what you wear (no classes) No item teleportation This worked: with a radical promises we got a lot of attention and by delivering on them in Alpha we built our early access program
- 27. SCALING WITH SUCCESS Original production value and scope were based on budget Early Access success lead to an increase in scope and production value This lead to further growth In the end achieved 3-4x the original budget production value through early access until launch
- 28. WILD GROWTH Wrote a novel with british author Peter Newman Novel flopped, but the story development was very valueable to guide content production Orchestral Soundtrack We also run our own regular AlbionTV stream which follows the developing player
- 30. MARKETING & PUBLISHING YOUR OWN MMO Marketing Youtube, Twitch, High quality preview videos Distribution Outsourced web development expensive & made integration harder
- 31. LAUNCH BILD
- 32. TECH FAILS #1 at one point, we had all buildings disappear in a closed alpha... ... which turned out to be a Cassandra bug (secondary index lost) "bleeding edge" open source technology is dangerous. Even release builds can have catastrophic bugs Do long-term testing before phasing a new middleware build into production! Consider using "older" builds where the bugs are already known!
- 33. TECH FAILS #2 (A POST-LAUNCH DISASTER) After 12-18 hours under full load, Cassandra suddenly became unresponsive for 5-8 minutes, killing the game Unable to fix this for > 2 weeks solution: get a consultant move from Windows to Linux (like everyone else) problem was disc IO bottleneck between Cassandra and Windows RAID controller driver Try to use technology the same way everyone else is using it!
- 34. DDOS ATTACKS Researched DDos protection ahead of launch Very expensive Technologically complicated (requires own AS, own IP range, BGP routers) takes months to implement so lets not do this
- 35. DDOS ATTACKS Got blackmailed and attacked > 4 weeks, > 60 Gbps attacks Sleepless nights, upset community burned through first mitigation provider (cheap is sometimes too cheap) Then got (expensive) external help Should have done this earlier!
- 36. INTERMISSION: DDOS MITIGATION Game Server Game Server Game Server Router THE INTERNET Criminals live here (customers, too) Our Data Center Public IP Range Clean Traffic Attack Traffic ReturnTraffic
- 37. INTERMISSION: DDOS MITIGATION Game Server Game Server Game Server Router Router THE INTERNET Criminals live here (customers, too) Scrubbing Center Attack Traffic gets filtered out Our Data Center Anti-DDos- Provider Public IP Range Clean Traffic Attack Traffic Clean Traffic Delivered via GRE Tunnel Redundant Routers (Fail-Over) ReturnTraffic Routers switch Public IP Range between our DC and Scrubbing Center via BGP protocol
- 38. BUSINESS LESSONS When we launched and sold hundreds of thousands of games, we were a company of almost 60 employees 6 months later we had to make the decision to scale the team to the actual player base Learning: When scaling up a niche game, remember youre still in a niche Just because a lot of people project their dreams into your MMO doesnt mean they will play it- those people you designed the game for will play it.
- 40. THE BETA FACTOR All throughout development we were discussing how big the impact of the game being in beta would be on player retention The argument was that players would not grind during a beta if they knew progress would be wiped Learning: Live player curve behaved very similar to Beta player curve
- 41. DESIGN LESSONS Surprisingly successful with our unemotional design approach following game theory and market logic Free Market & Game Theory works until it gets emotional. Personal design lesson: WoWification isnt everything.
- 42. CONCLUSIONS Niche MMOs can work! Early access can help you to build a much better product than you can originally afford Dont get carried away by your own success Middleware and clever use of tech is good, but dont ever cut DDOS protection on an MMO
- 43. THANK YOU! Questions? David Salz (david@sandbox-interactive.com) Robin Henkys (robin@sandbox-interactive.com)