SSDs I 1-2-1
£ Allowing for overhead IOPs (vMotion, cloning etc) an I/O requirement of 26,500 IOPs was agreed. Due to the efficient nature of VDI only 2.5TB of actual space would be required.
£ As an existing IBM storage user to provide the required IOPs in a traditional storage system they would need 180 SAS disks (each SAS disk is good for 150 IOPs). 180 disks adds up to 12 trays and depending on the vendor this could be 40 - 48Uand anywhere from 7-10KW of power and cooling requirement (depending on controllers exact configuration).
S3’s chosen solution was based on a single array with 18 x 200GB SSD drives. This configuration gives 3.6TB and 40,000 IOPs in only 2U (3.4”) and uses less than 500 watts of power.
Q Why is there such an interest in Solid State Disks (SSDs) right now?
MK: With SSD technology becoming more affordable and resilient, the storage industry has seen increased interest and investment in bringing to market new & innovative storage solutions that harness the superior performance, smaller form factors and lower power consumption of SSD compared to that of traditional HDDs. High profile acquisitions, successful company IPOs and the growing number of published case
studies, illustrating the significant increase in application performance and ROI, has gained the attention and confidence of a rapidly growing number of companies and organisations looking to meet performance, cost control and ‘Green’ requirements.
GW: SSD’s have been around for at least 25 years (I developed a DRAM based device in the late 1980’s, for example), but it’s only in the last few years that flash-based devices lowered the price points to a level that drove them into the mainstream market. However, the performance and reliability of flash-based devices have only recently improved to the point that they are acceptable to the professional IT practitioner. Also noteworthy is that flash technology is improving in all dimensions, at a far higher rate than spinning media - it’s already cheaper per IOPS, less power per IOPS, nearly as reliable, and closing in on the cost/GB though this last metric will take some years to achieve. The biggest remaining hurdles for flash are cost/GB, write-cycle durability, volatility, and consistent performance.
JC: With the proliferation of high IO demands - especially those wrapped around VDI and virtualization as a whole, solid-state technology is the natural fit.
Q And why so many start-ups in this space - what USPs are they all hoping to bring to market?
MK: NAND Flash technology has produced a paradigm shift in terms of increased performance capability; however, the storage capacities are still relatively low and come at a premium in terms of acquisition cost/ GB. This has lead to a number of start-up companies developing NAND Flash solutions, such as SSDs and PCIe based storage companies aimed at delivering maximum performance to critical business applications, as well as companies developing shared storage and caching technologies that maximise the efficient use of SSD or PCIe storage resources and provide greater levels of High Availabilty (HA).
GW: The “gold rush” is caused by people seeing the obvious - that SSD’s have some clear
business advantages, particularly in the transactional space. The start-ups think they have techniques to work around the remaining SSD weaknesses. Of course, it’s definitely an uphill battle as a start-up to try to sell
expensive hardware to the transactional, enterprise customers. If you look up “risk adverse” in a dictionary there’s a picture of an Oracle admin. Take Nexsan for instance - we’re over 12 years old and it’s only been in recent years that we’ve been taken seriously enough to be used as primary storage in a glass house data center. JC: A lot of this space was born out of the VC investments
WWW.SNSEUROPE.COM | SEPTEMBER 2011 21
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44