The High Availability Poll
-
Comments posted to this topic are about the item The High Availability Poll
-
Working in the legal industry, I can say that there is both a desire and the resources to make HA real. The last firm I worked for runs 2 large 2-node active/passive clusters. One of which was entirely dedicated to the billing system. The other mostly for the DMS but also sharepoint. Towards the end of my time there, they started to implement an off-site HA capability. Basically a simple fail over for the Blackberry database and Exchange server to ensure email would continue.
The firm I currently work for have gone entirely offsite. All SQL servers are running in virtual instances. The SANs are diffed off-site with equal hardware running the same VMs. Downtime for a complete fail over is a few minutes. Because of this, there is no other HA implemented. No clustering or mirroring or replication.
-
Desperately important AFTER an incident occurs that involves lack of availability.
-
Working in the energy industry, the business units don't require us to be able to recover instantly with HA options.
The fact is that this isn't a business continutity decision but a decision based on costs of the hardware and software.
It has been a real pain for me to convince the business units to take a good look at the possibilities and make a decision for the business continuity.
No success so far.
I guess the shit really has to hit the fan before decisions like this are made based on the right reasons.
-
I will happily join the cynics here and say we will implement HA after serious costly downtime. It doesn't seem to be terribly likely as our team have rarely lost anything important for any length of time. In this case, I think you need high penalties for an hour or so of outage (which you might otherwise occasionally encounter) or not be adequately organised to recover quickly.
-
Highly available database is a must for us (a large eshop). We used failover clustering but are going to move to AlwaysOn in a few weeks.
-
It is very important to us or my employer. I work for a large insurance company that is active in Scandinavia, the northern Europe. The department i work for is responsible for the large customers and it's very important that they can create insurances on the fly and have our support and knowledge available. We want our customers to succeed, that also is good for our business.
-
In the Lottery industry, 100% availability is paramount. Our company has likely invested as much time and money on HA database and system functionality as it has on actual software development. Noone wants to hear "Sorry, the Lottery is down" when there's a $100 million jackpot on the "line". 🙂
-
We have hardware up and running with databases all setup and configured at out DR site. In the event of a disaster we'd recover the dbs from our de-dup backups at that site. Our system owners are OK with a few hours of downtime in order to recover the dbs as they are not that critical to our business. I don't think I could ever convince senior mgt to spend the big extra cost of Enterprise Edition to get Always On. We are just now going to SQL2008R2 for these databases so we probably won't even go to SQL2012 and just skip it and go with SQL2015 anyway.
-
Yes, my company sees it as critical for both HA and DR.
We use a combination of technologies which include clustering and scaled out replication for HA and other technologies for DR.
The biggest thing to determine in all this is the TRUE need and the amount of time that you can be offline and then engineer a solution to meet those requirements. There are a myriad of options, even some homegrown that are better than nothing. Sure, your downtime might be longer with some of those options but that may be totally acceptable in the industry that you are in.
In reality all companies have a need. And there are always ways to get a solution in place, even on a very limited budget. We as DBA's just need to convince them to invest the time in getting something in place. 😀
David
@SQLTentmaker“He is no fool who gives what he cannot keep to gain that which he cannot lose” - Jim Elliot
-
It depends on what application the server is running. I work at a hospital so some apps are extremely critical and are set up in an HA configuration. Others, like our recipe database, aren't terribly critical and can be down for a while. Every department does have downtime procedures so if the computer system is down they can still take care of patients. It's just not as smooth and there's cleanup to be done afterwards.
-
cfradenburg (6/29/2012)
It depends on what application the server is running. I work at a hospital so some apps are extremely critical and are set up in an HA configuration. Others, like our recipe database, aren't terribly critical and can be down for a while. Every department does have downtime procedures so if the computer system is down they can still take care of patients. It's just not as smooth and there's cleanup to be done afterwards.I agree, it depends on the application. Some are point in time H/A critical, others tend to be pretty static and can be just restored to last nights backup without causing a lot of concern.:-D
"Technology is a weird thing. It brings you great gifts with one hand, and it stabs you in the back with the other. ...:-D"
-
I wonder if The Royal Bank of Scotland have re-evalutated any of their IT capabilities recently
- HA/DR
- back up/recovery
- Implementation/Rollback
:hehe:
-
The company I do most of my work for DO want the utopian 100% availability, but will never have the budget or staff to implement and run it. We operate a real-time telephony platform, and any failure has the potential to cause significant disruption. Directors won't shell out a large sum of money for a solution, only to be told there will still inevitably be downtime when an incident occurs. There are many processes that still require manual intervention to bring back on line, and also data to be checked for potential inconsistencies before declaring the platform live again.
A recent blue screen on one of our blade servers (caused by a bad memory module) resulted in an ASR restart - the database came up OK, but there were still other parts of the system that needed to be restarted manually for a variety of reasons. So we largely rely on continuous monitoring, SNMP, and documented procedures to follow when something looks like it has failed.
Back in "the olden days" (i.e. about 5 years ago) we implemented a failover solution based on LifeKeeper, mainly because it did not require Enterprise software or clustering, but also because it "protected" more than just SQL Server. The problem was that nearly every incident we ever had with database availability being lost was not detected by the LifeKeeper watchers, and so we were no better off. That experience has made people very wary of systems that claim to keep the whole world turning whatever the failure.
In the end it is down to budget and expectations, and unfortunately the business climate is such that the level of investment required is unlikely to be found when the resultant system still isn't bomb-proof.
-
I've been working at a rapidly growing logistics company for the last year where most most of the operational databases are distributed across the continent. It ain't pretty and I'm trying to overcome the small shop mentality. We are centralizing into an off-site data center soon, but most still don't understand the difference between HA and DR. They are surprisingly tolerant of availability outages if they aren't more than an hour or two. But I don't expect it to last. The cost and complexity of HA seems to be little understood. Outside of the financial sector (in my experience anyway), there is a perception/expectation that HA is a plug-in magic bullet and that it is primarily a question of hardware outlays. Convincing the powers-that-be to spend the necessary money can be a real challenge, but if you want something even more challenging try to convince them that with a little bit of process discipline they can achieve some degree of HA with little cost.
Viewing 15 posts - 1 through 15 (of 20 total)
You must be logged in to reply to this topic. Login to reply