|
|
| Author |
Thread Statistics | Show CCP posts - 6 post(s) |
Dr Slaughter
Coreli Corporation
Corelum Syndicate
   |
Posted - 2007.08.03 22:23:00 -
[1]
Edited by: Dr Slaughter on 03/08/2007 22:23:52
Ran the technical services for a systems management vendor for 5 years in Northern Europe which worked exclusively in the MS server space. Speaker at TechEd etc. Implemented MS server products worldwide, for the likes of CSFB, Shell, etc. etc. Implemented SQL clusters. Wrote software to monitor and manage SQL blah blah. Current company provides systems management consulting, is a gold partner, and RDP partner...
background check over.
I would highly recommend the following tools to help if you don't have them already:
Idera's SQL Diagnostic Manager & SQL Safe (down time each day could possibly be shorter). While we're at it you might want to use some of their other products too like Change Manager, Schedule and Config.
On a different topic but of equal interest to your DBAs and storage people might be SonaSafe for SQL Server.
One of my businesses is a distributor of both products.
Oh, and the person who said MS should be helping you out is dam right. Hopefully you have a Technical Account Manager (TAM) who can get some one from Microsoft Consulting in to help.
Dr S
ps. Wrangler you're welcome to drop me an email if you want to swap notes with one of our SQL consultants or see if we can help you guys out.
 Originally by: fire 59
Arguing with stupid people is like trying to bite your own elbow.
|
Dr Slaughter
Coreli Corporation
Corelum Syndicate
|
Posted - 2007.08.04 14:08:00 -
[2]
Originally by: zombu2
i won t support M$ whatsoever
Then perhaps you should stop playing a game who's developers use a huge amount of Microsoft technology, by which, you are indirectly supporting the M$ (-sic-) revenue engine.
Originally by: fire 59
Arguing with stupid people is like trying to bite your own elbow.
|
Dr Slaughter
Coreli Corporation
Corelum Syndicate
|
Posted - 2007.08.04 16:10:00 -
[3]
The debate about which SQL system to use is utterly pointless.
CCP picked MS SQL.
Moving from MS SQL to something else isn't about to happen within the time frame this problem needs resolving.
They've had 'issues' with their database for years and people, myself included, have provided feedback on what might help fix the problems, or better manage their platform.
I'm sure they've taken some of the advice and acted on it, I'm also pretty sure they haven't acted on lots of it too.
Let's face it they're a game company, not a bank, and they probably didn't have the money to go out and buy a OS/400 based solution (or similar) at the start and the cost of migration would be huge.
CCP, please, just look at implementing some decent systems management tools if you haven't already and if you have but need help getting more out of them let us know. Using something like NetIQ's AppManager or Microsoft Operations Manager, or even a little bit of programming and scripting it's possible to monitor MS SQL server in minute detail.
My company would be happy to quote you for implementing a solution and if needed help the developers instrument their code to better support health, performance, and availability management.
It really should be pretty easy to get you set-up to be informed that something is going wrong before the server node actually goes POP.
Couple of questions I'm curious about:
What systems management solution have you deployed?
Is the Cluster running active-active or passive-active?
How long is the fail-over taking (on a good day)?
What sort of delay can the proxies deal with before their clients (us) start having problems?
Do you have a 'dummy' proxy (i.e. monitoring service/script) which gives you end-to-end performance figures, or have you implemented some management instrumentation into the proxies themselves?
How are you writing your stored procedures and triggers? Is it mainly with the .NET CLR or T-SQL?
Originally by: fire 59
Arguing with stupid people is like trying to bite your own elbow.
|
Dr Slaughter
Coreli Corporation
Corelum Syndicate
|
Posted - 2007.08.04 16:25:00 -
[4]
Originally by: ritters****2
Originally by: Angua Forres
Whenever I hear "the [one?] SQL server failed" I get an ugly fealing that the architecture is too much central-focused for their own good.
Basically wit MS SQL you can have up to 8 nodes in your cluster but a database instance can only run on one node at a time.
Some people deploy multiple instances, say with 2 nodes, each with their own instance, and that's referred to as active-active but each instance doesn't know about the other. So, to be honest, unless you can have multiple instances and spread the client proxies across them, they're pretty much stuck with only having the db run on one node of the cluster at a time.
My guess is that at the moment when everything starts to go wrong, the time it takes to fail over is enough to cause the client proxies problems... and poof the rest of the castle comes crashing down.
Originally by: fire 59
Arguing with stupid people is like trying to bite your own elbow.
|
Dr Slaughter
Coreli Corporation
Corelum Syndicate
|
Posted - 2007.08.04 16:39:00 -
[5]
Originally by: RapidTaco
IF ANYONE FROM MICROSOFT IS READING THIS FIRE UP THE RDP OR GET ON A PLANE!
or we could just all bug them, goon style, at: their blogs
Originally by: fire 59
Arguing with stupid people is like trying to bite your own elbow.
|
Dr Slaughter
Coreli Corporation
Corelum Syndicate
|
Posted - 2007.08.06 15:11:00 -
[6]
Originally by: Panshun
Edited by: Panshun on 06/08/2007 11:41:44(yep, I've quit IT now, feels good)
   Lucky sod.
Pity they don't seem to be able to put the market onto it's own instance and effectively go active/active on the db cluster.
Originally by: Red Button
Someday (hopefully soon. 30/08/2003) load balancing across SOL servers will be fluent and load will shift between servers as the service loa |
|
|
| |
|