Whiptail whips out SME-friendly flash array Enterprise flash array fettler Whiptail has announced a flash array for the rest of us: the WT-1100, aimed at branch offices and small/medium enterprises. It's a diminutive 1UK rack enclosure holding up to 4TB of flash which delivers 100,000 IOPS with a latency less than 0.1msecs. The device runs the same RACERUNNER software …
(from the article) - is it shitty IOmeter 100% read? Usually what folks like that try to pimp.
Fenwick - I think the obvious answer to your question is that commodity crap is just a server with a bunch of SSDs, with no redundancy whatsoever(nor software features - thin provisioning, dedupe, compression, replication, etc), no support, no flexibility, and an truly unholy number of single points of failure.
If you have a workload that is well suited for such a system - then by all means have at it
For those that think ZFS might fix a lot of those issues, think harder. Or google "zfs availability vs reliability" and you can see an excellent mailing list post I quoted on my blog a few years back on that specific topic. I couldn't of said it better myself.
Really if you find yourself seriously asking that kind of question your better off letting other people do the technical work.
I guess you answered my questions. You pay for software and support. I should have asked: What is the difference in the hardware?
Yes, I know that on the SI quality scale the box I mentioned was "commodity crap", I know that it is set up in a crazy way, with a reliability rating of "unholy". Looking at their website, I don't know that the whiptail box isn't just the same hardware with an "enterprise" badge stuck on the front.
Is it? If not what is it? Why is the whiptail hardware better than the commodity server? Why is the ACCELA hardware better than 10 (for redundancy) commodity servers? (or if not whiptail some other similarly expensive SSD based device) Is the entire price difference due to the software, support and high margins?
And finally yes you're probably right. Being new at this stuff, I shouldn't ask questions because I might expose how ignorant I am.
1U : Where are the redundant controllers for high availability?
100K IOPS : IOmeter results, thats is probably 50K in real-world testing
20,000: That is $5/GB for hardware alone. What about the cost of support and maintenance.
What about developing new technology like the rest of competition to reduce flash cost and offer more value to customers: What about In-line deduplication and compression?
Is this the best you could deliver with the $31M you received in your C-round http://www.theregister.co.uk/Design/graphics/icons/comment/wtf_32.png
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
