Player Features and Ideas Discussion

Not sure how to start this out, so I guess I"ll start with some background.

From what I've read, CCP seems to be very heavily server based technology, lots of hardware, lots of power consumption, create a series of nodes. I was wondering if CCP fully investigated the idea of switching to a mainframe based solution for the eve world / environment?

I've worked with banks, and e-discovery offices before, and have completely virtualized massive server based environments, not only improving performance, but increasing stability. It seems that a virtual environment with the timezone / way the player base moves around is absolutely ideal to this system. I'm not trying to tell anyone at CCP what to do, I'm simply raising a suggestion that many fortune 500 companies run (and isn't nearly as expensive as most vendors such as HP etc, try to tell you it is)

First off, allocating processing power where needed:

Imaging if you have 500 servers, running the eve universe, separated by clusters, or however eve does it. Say Russian's are on in drone region. You have tons of servers, running at say only 1% of there total capacity in certain regions of US timezone held space and are inactive, but you can't use that CPU potential elsewhere. As a side note if a server is only running at 1% of its total capacity it is still sucking up the same relative amount of power as if it were running at 100% of capacity

Now, during this time, a fleet fight occurs out in drone region space, and time dilation kicks into effect (great solution btw). You are still limited from all those people being stuck on the SAME server (limited by the number of transactions the server can handle per minute). Even if you have a dedicated server for the particular system, the load of say 1100 people fighting in the same system, is simply too much for that server to handle, thus, transactions gets lots / ignored, or your time dilation goes into effect (once implemented)

Now lets look at a cloud based environment. In a cloud base, you can still create virtual clients to run with the same server code eve has now, simply on virtual servers. So the transition to a mainframe / cloud-based environment isn't one that completely destroys any specific coding between the current clusters / coding style that eve has.

Say, there is a fleet fight over in Catch. 4000 pilots, actively engaging a system / defending a system timer. Say you are one lonely pilot flying around on the other side of null sec with no other pilots in system. If its a cloud based system, it takes the unused potential of other sectors of eve, and can apply it to the fight. (you can then have the processing power of 50servers, for one massive fight!)

What does that mean? Well, first off, Mainframes can be upgraded to have the potential of thousands of servers, not just hundreds. They are cheaper to maintain power wise, and can run massive cloud based environment. Plus, the potential is huge. In theory, these monsters that run banking centers, many defense departments, and NSA signal intercept / decryption, could be utilized to expand eve to incredible horizons.

No more like in 10,000 man fleet fights, no lag, no lost transactions, simply pilots fighting it out. The possibilities that eve could accomplish with this technology is endless.

Int he end I'm not trying to tell CCP what to do, just simply pointing out a solution that may work for you guys. I hope you pass this on to your techie geeks and tell them to seriously look at and consider this. I took a company running 200 servers, to 1 mainframe, saved them 100k dollars a year in POWER alone, increased there efficiency / speed of there product, and the one I know everyone hates, reduced the amount of tech's necessary to run / manage the network and equipment.

I hope you guys consider this as you continue to make improvements on EVE and your hardware in the future.

Don't think the word 'mainframe' is in much use anymore, nowadays they call it 'the cloud' or if we go back a couple years, clusters.

Most modern mainframe design is not so much defined by single task computational speed, typically defined as MIPS rate or FLOPS in the case of floating point calculations, as much as by their redundant internal engineering and resulting high reliability and security, extensive input-output facilities, strict backward compatibility with older software, and high hardware and computational utilization rates to support massive throughput. These machines often run for years without interruption, with repairs and hardware upgrades taking place during normal operation - Wikipedia.

I'd say CCP is doing *exactly* this. Afaik they use IBM's BladeCenter 'mainframes' jammed into a few racks somewhere in London UK. The exact design of the software is largely a huge unknown to the public, except on key technology choices. Like RAMSAN's and running the database on MSSQL.. etc etc. There might be more detail in some old dev blogs if you search around...