What are the pros and cons of developing your own game engine?

Question

Game development

I say right away, I was on Google, the articles are all old + do not give an answer to my question. Do not offer thematic forums, there are too many trolls.
We make the game MMOFPS. The scale is large, with 5,000+ players expected at launch.
None of the ready-made engines suits us (we looked at BigWorld but their support is silent + negative feedback from those who use it, CryEngine - again dead support + weak server part + it is not clear how to be friends with the web part). It was decided to write the engine ourselves (there are no programmers yet, they, by the way, are the rest of the team that they have not yet found), so far I see the following pros and cons:
+great flexibility (we write what we want in the most rational way)
+security (yes, a lot of security depends on the developers. But it's still safer than shoving crutches into someone else's finished product)
+cheaper (I understand that writing it yourself is also not free, but fork $ 1,000,000 for the engine + a considerable amount of money for programmers for rework will be more expensive)
-long (although again, not so long than ready + rework. Yes, and you can use time wisely)
I found out the reason for the "dead" support, they do not want to deal with our German LLCs or USA give

Answer 1

Unity3d client + apache thrift network + own server in any language. Enough to get started. And there, on the final version, rewrite thrift rpc to your network.

Answer 2

Pros: experience
Cons:
- but experience is much less than when using a good engine.
- the result is much worse: both opportunities, and stability, and support.
- VERY long time to write.
Now even AAA games are almost all without exception on non-self-written engines.

+ security (yes, a big measure of security depends on the developers. But it's still
safer than shoving crutches into someone's finished product)

Foreign engines are time-tested and a large number of pros. They will definitely be more reliable.

+ cheaper (I understand that writing it yourself is also not free, but paying $ 1,000,000 for the engine + a considerable amount for programmers for rework will be more expensive)

Themselves much more expensive will come out if you need a similar quality.
And many engines are completely free with small restrictions: Unity, UE.

Answer 3

I have three release projects and each has its own 3D engine. I will voice my thoughts about
Self-written engines are justified if:
1. You have prehistoric or specific hardware or platform
2. You have very specific visualization tasks
3. You have a local non-game project, resource consumption is important to you
In any game projects focused on mobile and themes more desktop platforms - you need to use existing (more than three) ready-made frameworks. Unity3D - one of the best options, speed, cost, entry threshold.
And about "experience". Gaining experience when writing your own engine (and this is not just a render, if not) is a rather controversial achievement, negative experience is also experience, but spending effort, money and time on “we did everything wrong” is an unaffordable luxury.

Answer 4

In fact, they have already answered, but I will also add my 5 kopecks (for the same rake).
Java+Apache mina server part, MySQL base, XML content (point by point - cross-platform, free, fast, relatively safe, portable to С/SQL without any problems).
Client - Unity3D (for the time being, the development is free, so far the necessary funds are enough)
Writing your own engine == hopelessly behind the times (unless, of course, you don’t have a staff of shark programmers who have been involved in game development for several years). To be convinced of this, it is enough just to figure out a plan for the implementation of some thread of the "elite" feature of modern games - for example, water. Appearance, continuity, reflections, glare, post-effects, immersion...
I agree about - "they did not write a single game, but swung at MMOs" - writing MMOs, and even FPS, is a non-trivial task. Have you looked at the article describing CounterStrike's ping compensation system for shooting clients? And there are many more nuances - displaying targets in real time (so that they don’t look through walls), protection from head-aim bots, movement control (a player running at a speed of 20 km / h is a speed hack? And if he falls?). You are now choosing an engine for the client - but you didn’t say a word about the server part - and this is 70 percent of the entire project: no matter how beautiful the picture is, but if the player constantly crashes / falls through the textures / gets banned for an imaginary speedhack - the game will not popular.
Nevertheless - if you need a ready-made constructor - I recommend Unity3D
If there is a handy proger - Ogre3D or AnarchySDK
Well, for a startup - PureBasic (if you smile now - look at Yandex.pictures of the game on it, with photon and ngui)
In any case, a good start.

Answer 5

Greater flexibility - yes, adjusted for two to three years of pre-development. Why are you not satisfied with the flexibility of CryEngine or Unity? They just don't do anything.
Security - are we talking about the graphics engine? Are you afraid that someone will steal the textures? If you are talking about interaction with the server, then no one will give you a ready-made system out of the box anyway, you will have to cut it for yourself anyway. And then test for a long time, because everyone has vulnerabilities.
Cheaper - and who will write the engine for free? Unity is worth a penny compared to developing your own engine.
Long is not the right word. In a couple of months you will have only static models with primitive lighting, then you will still have to make a system of materials, animations, shadows, post-effects ... And it will only work on a limited amount of hardware. And fall due to some strange features of the drivers in unpredictable situations. Your engine is sooooo expensive. Building your own engine is always a bad idea unless you're Ubisoft or EA.
For some reason it seems to me that you have not yet made a single game, and are already undertaking to make an MMO. Start with something smaller, or at least make a prototype on an existing engine. And then you solve the problem of five thousand players, and you not only don’t have a game, you haven’t even found people who will make it.

Answer 6

+ cheaper (I understand that writing it yourself is also not free, but paying $ 1,000,000 for the engine + a considerable amount for programmers for rework will be more expensive)

It will be cheaper if only you write a much simpler and highly specialized likeness of these engines.
Are you sure that you have bothered to study the capabilities of the existing engines sufficiently to draw such conclusions?