Faith In Technology

krashkart · 03-13-10, 04:45 PM

I just came back in from the SH5 forums, feeling a little sweaty. It's hot in there. After seeing the topics about Ubi's servers being down I wonder (yet again) about the future of online services - the games, applications, so on. All it really takes to muck up a perfectly good idea is a stiff bitchslapping of some kind. Be it DDoS, the servers blowing chunks, or what have you.

Given the recent failings of a particular DRM model *cough*, I wonder what solutions the Subsim community thinks would harden those kinds of services, and improve their reliability.

I believe, first of all, that aforementioned company deployed their scheme a tad early. Additionally, I feel that a redundant backup would have helped ease the pain.

Please note: This Is Not Intended To Be An SH5 Rant Thread - Please post SH5 rants in the SH5 forum

What, if anything, would keep those services up and running in a whirling sh!tstorm?

TarJak · 03-14-10, 03:37 AM

The only thing that will keep this sort of thing running in a traffic storm is a scalable architecture that either has no bottlenecks or protects any existing bottlenecks from large volumes of traffic in some way.

The most common problem with this sort of system is usually the capacity of the backend databases to deal with the calls being made on it, particularly if it is being called on to do heavy duty read-write operations.

Web server scalability is relatively easy particularly if you can afford services like Akamai which uses a distributed content network to allow content to be cached. As soon as you introduce a database that needs to be updated in any way, you introduce the possiblity of a bottleneck unless you have a nice way of having a distributed database within the content network designed in such away that the synchronisation of the data cannot impact the performance of the system.

It is tricky stuff to do but is possible if you have the money and access to the right smarts to make it work.

03-13-10, 04:45 PM	#1
krashkart Navy Seal Join Date: Jan 2010 Posts: 5,292 Downloads: 100 Uploads: 0	Faith In Technology I just came back in from the SH5 forums, feeling a little sweaty. It's hot in there. After seeing the topics about Ubi's servers being down I wonder (yet again) about the future of online services - the games, applications, so on. All it really takes to muck up a perfectly good idea is a stiff bitchslapping of some kind. Be it DDoS, the servers blowing chunks, or what have you. Given the recent failings of a particular DRM model cough, I wonder what solutions the Subsim community thinks would harden those kinds of services, and improve their reliability. I believe, first of all, that aforementioned company deployed their scheme a tad early. Additionally, I feel that a redundant backup would have helped ease the pain. Please note: This Is Not Intended To Be An SH5 Rant Thread - Please post SH5 rants in the SH5 forum What, if anything, would keep those services up and running in a whirling sh!tstorm? __________________ sent from my fingertips using a cheap keyboard

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

03-14-10, 03:37 AM	#2
TarJak Fleet Admiral Join Date: Jul 2006 Location: Sydney, Australia Posts: 17,052 Downloads: 150 Uploads: 8	The only thing that will keep this sort of thing running in a traffic storm is a scalable architecture that either has no bottlenecks or protects any existing bottlenecks from large volumes of traffic in some way. The most common problem with this sort of system is usually the capacity of the backend databases to deal with the calls being made on it, particularly if it is being called on to do heavy duty read-write operations. Web server scalability is relatively easy particularly if you can afford services like Akamai which uses a distributed content network to allow content to be cached. As soon as you introduce a database that needs to be updated in any way, you introduce the possiblity of a bottleneck unless you have a nice way of having a distributed database within the content network designed in such away that the synchronisation of the data cannot impact the performance of the system. It is tricky stuff to do but is possible if you have the money and access to the right smarts to make it work.