If you truly try to push your IO subsystems to the limit, a Fusion-IO subsystem will expose more weaknesses than other storage subsystems because it has so much more throughput. Buyers need to be aware that this is a version 1 product with version 1 best practices and documentation. In my post about adding reliability to your infrastructure , I pointed out:. The faster we wanted to go, the more careful the team had to be with other pieces of the infrastructure.
FusionIO drives solve a real problem, and they can deliver staggering performance, but just like any other new technology, you should test them thoroughly in your own environment before deploying them in production.
In the case of Fusion-IO drives, you should probably even test using similar power supplies to production in order to improve your odds.
You could argue that this provides a higher availability than clustering, since it uses two different storage back ends. I wish I had SQL Server benchmarks for these scenarios to share with you, but the testing process ended up taking several weeks, and I ran out of time.
My verdict: these drives can solve some tough challenges. Quite the opposite — I sank a lot of time into this project. Folks who follow me on Twitter may remember that I struggled with Fusion-IO during the initial stages of the testing as we went through one hardware failure after another. Their throughput may push the rest of your infrastructure to its limits, but hey — that can be a good problem to have!
If you liked this post, head over there to see what else is happening in the blogosphere this week! Your review leaves me to see some great options for read only partitions as well. That is if they are small enough to fit on the drives. This review will help me formulate some great questions for them today. Nice write up, Brent. I might disagree about the prevalence of synchronous database mirroring. I know of several much bigger customers are using it also.
Out of all your SQL Server databases, what percentage use synchronous mirroring? We have done both synchronous and asynchronous mirroring at NewsGator. We like it because how easy it makes rolling upgrades with a second failover , and because it gives us two copies of the data, above and beyond our backups.
Glenn, I am interested to hear more about how you guys use Sync. I have seen lots of Async use but never seen a production system that uses synchronous.
Run a cluster for high availability, then do synchronous mirroring to another server in the same datacenter or extremely nearby. You still need another solution for disaster recovery like log shipping or SAN replication to a disaster recovery datacenter, though.
Great review Brent! I also recently helped a customer Setup Mirroring for a fail over situation going from one data center to another data center with mirroring on One data center just recently failed as well so they were able to test the solution and it worked well for them.
Question — in your testing did you ever sustain a test where one or both of the drives filled up? What happened then? Can you elaborate on that? I am the Principal Solutions Architect at Fusion-io and would like to start by thanking Brent for posting such an objective review of our technology.
This is indeed very useful to the SQL Server community. I am also grateful to Brent for the patience he showed us while testing our product. I guess for Fusion-io memory the XFS numbers will be even worse. There are several card models. SLC sustains more writes. MLC is good enough for normal use.
Knowledge Base Contact Login Search. Membership is free, and your security and privacy remain protected. View our privacy policy before signing up. Follow Us Twitter LinkedIn.
Join the newsletter! Sign in with LinkedIn Sign in with Facebook. Autonomy's former chief financial officer has alleged the firm collapsed partly because two financial analysts agreed to badmouth it in the hope of making a profit from its demise. He accused former JP Morgan analyst Daud Khan and industry colleague Paul Morland, once of analyst firm Peel Hunt, of carrying out "concerted and improper efforts to depress Autonomy's share price for their and their short-selling clients' benefit" and being "vehemently antagonistic towards Autonomy and its management.
If you struggled to get into your Gmail this morning, it wasn't just you. Unhappy users from Europe all the way to South Africa reported a significant outage. The issues kicked off at around 8. The Register - Independent news and views for the tech community. Part of Situation Publishing. Review and manage your consent Here's an overview of our use of cookies, similar technologies and how to manage them.
Manage Cookie Preferences Necessary. Always active Read more These cookies are strictly necessary so that you can navigate the site as normal and use all features.
Sign in. Topics Security. Resources Whitepapers Webinars Newsletters. Get our Tech Resources. Next page: What does Woz do? Share Copy. Corrections Send us news. Other stories you might like He called himself the King of Fraud. Microsoft engineer fixes enterprise-level Chromium bug students could exploit to cheat in online tests Ability to block 'view source' for specific URLs hasn't actually worked for years.
Apple is beginning to undo decades of Intel, x86 dominance in PC market What a difference a year makes? Still feeling secure? Microsoft admits Samsung phones under Intune mobile device management are dropping out of compliance Auto-restart or manual update requires manual interaction to bring a gadget back to the light.
0コメント