Post
by Spamoni » Tue Oct 01, 2013 6:51 pm
This is the most insightful post I've seen regarding the GTA:Online issues
Hi all,
I know that we're all frustrated with Online - and the general consensus is mudslinging at R*... I'm an infrastructure engineer myself - and build out server environments myself (albeit for apps for corporate enterprises rather than games). I've been lurking on here for a little while but finally joined as I think I can offer a perspective that is often overlooked.
Sure - what I have to say won't fix the problem. It won't make everything suddenly work - and it won't answer everyone's questions or make people feel better... But I'm guessing understanding what's going on a bit does help.
So - one of the big criticisms of R* is that they "should have been prepared" and "they've had an extra two weeks since launch", etc, etc. I hear ya - but what you have to remember is that R* is a business and they can't have enough capacity for every game sold to all play at once for the next 10 years. If that were to happen, then each game would probably cost in the region of $500 - $1000 a copy.
The simple fact is that only some people will play the game, and for various reasons - time zones, work / school schedules, eating habits, family and other commitments - not everyone will be playing at once. So... R* needs to provide enough capacity with servers - and just as importantly bandwidth to those servers - to enable people to play.
In the first few hours and days, we're all clambering to play as early as possible - heck, we've already had to wait an extra few weeks, right? The thing is that there's more people trying to play now then there will normally be. What should R* pay for? Should they pay for capacity for everyone to play now and then have all that extra capacity afterwards when numbers level out? If so, that has to be paid for from somewhere.
Of course, hopefully R* will be utilising some form of "elastic" cloud servers and networking. Apologies for those of you that knows that this means that they can add more servers and bandwidth, etc "on demand" (I'm not trying to teach people to suck eggs, just set a level of understanding for everyone reading).
Whilst this allows on-demand upscaling / downscaling of servers available - it also means that disk space and the building out of templates to allow the quick provision of servers needs to be accounted for. If there's too much demand then new servers may need bringing online at a new datacenter, or where hardware hasn't got access to templates to build new servers. This means that extra time is needed to bring up new servers in new locations.
Also, if a larger number of servers are needed then it can sometimes have to wait until the datacenter has physically got the resources available to provide (I'd hope that this wouldn't happen though!) - and in some cases if an agreed threshold is surpassed it can mean that until a new Purchase Order is signed off by the customer (in this case R*) with the host/datacentre, new servers can't be deployed.
Also you have to remember that not everyone wants to play online - and so a huge guessing game would have had to have taken place with R*. Admittedly it would have been an educated guessing game, based on number of sales... But very little else!
Another of the criticisms being levelled at R* today is that people don't like being "beta testers" and that's what we are...
When any new service goes live - no matter how much testing has been done - there will always be issues. A testament to a company and the services they and their suppliers provide is how they react to those issues.
R* *could* have done a staggered roll-out - they could have made people sign up to R* Social Club, take their place in a queue and opened up to say a thousand people a day to keep capacity issues at bay. People would've complained about that though, and there would've still been issues that wouldn't be experienced until you had a critical mass of users.
Alternatively they've gone the other way - they've opened up to everyone. The game's sold beyond their wildest dreams. They knew it was gonna be popular - but they didn't expect it to be by this much. It's easy to say "oh, they've had two weeks in the know" - but the fact is that they haven't - the games continued to sell even between now and opening day. For all of the reasons above, I hope it's easy to see that you can't just bring enough capacity online within minutes.
Let me try and put an analogy on it. You invite a load of people to dinner. A few weeks out you have a pretty good idea of who's coming and who isn't, but you haven't quite heard from everyone. A few more people ask to come and you think you'll be okay so say yes. You then make sure you have enough food ordered, and know how you're going to cook it all. The evening before you get a phone call from one of your kids asking if another 3 people can come because they've got a power outage. You say yes, dash to the store only to find out that they're out of some of the food you need - but that there should be a delivery in the morning. The big day comes, but you have to wait for the store to open - and your guests have started arriving before the store takes delivery... What do you do?
For me - I'm frustrated like everyone else. But it's where we are in a couple of days that matters. If R* have it sorted, great... If not then I think it's fair to start asking bigger questions.
I can imagine those involved with infrastructure capacity at R*, along with engineers with their suppliers, are run off their feet, will be working very long hours today and for the next few days, and are doing their level best to get everything sorted for us all....
