Pages: 1 [2] :: one page |
|
Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Elvis Preslie
NRDS Securities Apocalypse Now.
25
|
Posted - 2013.04.08 15:00:00 -
[31] - Quote
uhmm do you guys think they love having downtime? downtime HAS to be there for MANY reasons.
The main reason is so that they can make a backup of everything; even though it could be incrementally backed up, imagine losing so much SP when one of their hard drives crash, just because you don't want downtime. Seriously, imagine finally obtaining that titan and the sp for it but one of the hard drives on the cluster crashes and they hadn't done a backup recently.
you cant make a full CONSISTENT backup of everything while it is being used or changed. Yes, you can use RAID mirroring and all sorts but imagine the chance of ALL The backup drives failing at the same time; nothing is fail safe EXCEPT for them backing up everything to TAPE and storing it in a Led vault.
The next reason is so they can apply patches and fixes; when are they suppose to do this if the cluster stays up all the time? It wiill always be an inconvenience for someone. Imagine being in the middle of a titan fight and being told "downtime will occur in 10 minutes". At least with downtime being how it is, EVERYONE knows when it is and when to not start a fight you cant finish.
When you have a solution for them, THEN make this petition. That's why most of my ideas get put in the game, because they're balanced and pheasible to program for CCP. I'm a developer but I've never done C++, only C and other non graphical. |
Draconus Lofwyr
EntroPrelatial Industria EntroPraetorian Aegis
41
|
Posted - 2013.04.10 22:00:00 -
[32] - Quote
Chelien Dallocort wrote:Draconus Lofwyr wrote:and if as i said earlier, they had programed the server code multi threaded, reinforced nodes would not be needed. Looks like we got an expert here, can you explain us why? You seem to be familiar with the implementation of the server side. A fact is that multi threaded code isn't a surety to be faster than single threaded code, the opposite can happen even with perfect implemented code.
Because i have been playing the game for long enough to remember CCP saying this was a major bottleneck with their code. multithreading allows for more cores or CPU's to handle the code at he same time, instead of being locked into one core, allowing for dynamic expansion as the load calls for it. Also, being in IT as a systems Admin, and a network engineer, and knowing enough about programing from dealing with a lot of programers in my Admin roles.
instead of cherry picking one statement out of many, you will see that I said it was a gamble, not poor planning, these are choices that had to be made, everyone make mistakes, its how they deal with it that makes the difference. CCP has done a remarkable job dealing with a mistake that is extremely hard to fix after the fact. consider it on the scale of replacing a jet engine, while in flight, while the passengers are watching. |
The Baby-sitter
Vitriol Ventures Tribal Band
2
|
Posted - 2013.04.10 22:21:00 -
[33] - Quote
Little Dragon Khamez wrote:Downtime is great! You got to have some time to change the intravenous drip, refill the coffee flasks, load up the food trays, get more stimulants and change your nappy.
I agree - there is nothing wrong with the current DT - It's only 30 mins instead of the hour we used to have. Suck it up man - go talk to your wife, kids girlfriend or go Fap or something.
Seriously I live in a TZ that is most affected and I still think its a minor compared to a crappy state of a semi broken server.
HTFU |
Mara Rinn
Cosmic Goo Convertor Cosmic Consortium
3231
|
Posted - 2013.04.10 22:46:00 -
[34] - Quote
Hoinus wrote:Xindi Kraid wrote:If my server status indicator is actually telling the truth, I have seen DTs of 10 minutes.
That's hardly any time at all. I still remember when it was fortuitous for the cluster to be back up within the hour. Again, its not the length. its the time before it, don't feel like doing anything major when its DT soon.
^ This.
No point running missions or exploration sites or any other activity with a long but not certain duration before DT. DT has been as short as 4 minutes lately, it's not the duration of the downtime that matters, it's the fact that it happens at all.
Dev blog on "Death to Downtimes" here: http://community.eveonline.com/news/dev-blogs/death-to-downtimes/
Day 0 advice for new players: Day 0 Advice for New Players |
seth Hendar
I love you miners
38
|
Posted - 2013.04.15 12:35:00 -
[35] - Quote
Chelien Dallocort wrote:Draconus Lofwyr wrote:and if as i said earlier, they had programed the server code multi threaded, reinforced nodes would not be needed. Looks like we got an expert here, can you explain us why? You seem to be familiar with the implementation of the server side. A fact is that multi threaded code isn't a surety to be faster than single threaded code, the opposite can happen even with perfect implemented code. a code made to be able to use multiple cores can be spreaded on the fly on additionnal cores.
if you have say 2 core / system as a regular basis, an increase in load in a system could then have as a response to dedicate more core to said system on-the-fly.
for me, the best answer, as the system is currently, would be to switch to virtual machine.
this would allow a VM to have more blades allowed to it dynamicaly, without requiring a rethink of the whole code, and thus the ressources to be shared more efficiently accross he board, this could even lead to TIDI being made not used anymore! |
Naomi Knight
Imperial Academy Amarr Empire
290
|
Posted - 2013.04.15 12:46:00 -
[36] - Quote
It should be moved 4 hours earlier. |
Anthar Thebess
REPUBLIKA ORLA C0VEN
89
|
Posted - 2013.04.15 13:42:00 -
[37] - Quote
Or simply move it to 02:00-03:00 EVE 3 days a week. |
|
|
|
Pages: 1 [2] :: one page |
First page | Previous page | Next page | Last page |