 |
| . |
"Greasing the dataflow through
legacy hard disk arrays" - by
Zsolt Kerekes,
editor StorageSearch.com
(this article was published here in
October 2009) |
The number of vendors
in the
rackmount SSD
category will rise to hundreds in
the next 3 years as enterprise users become more familiar with the benefits of
this type of storage.
Market entry costs are low...
It's easy for
RAID systems oems to
requalify their products to use SSDs. But unlike the transition to
NAS which we chronicled in
the late 1990s (where low performance products found valid application niches if
they were cheap enough) the role of the SSD array today is primarily server
acceleration. That means not all
hard disk boxes will be
optimal for use with SSDs. Slow bus interface adapters and slow
RAID controllers will
eat into the latency budget and squander potential SSD performance. I have no
doubt that in addition to many worthy products - some abysmal SSD arrays
will also come to market (populated by no / low performance SSDs) as floods
of lemming like companies leap into the enterprise SSD surf.
Looking
at future trends - the rackmount SSD market can be grouped in 4 main segments
- some element of open architecture RAM SSDs - like those from
Solid Access
Technologies, Third I/O
and (later) Kaminario
- in which industry standard servers are used as the platform to build the SSD
(using proprietary software).
I
think that all 4 product types will survive in the market for several years.
The
advantage of proprietary SSD rackmounts is they offer superior
performance compared to COTS arrays of flash / RAM SSDs.
Apart from
applications which need the ultimate performance - which can only be met by
proprietary architecture SSDs - there can also be applications in which the
proprietary products offer superior price / performance or reliability compared
to "open SSDs" - due to what I call - "big SSD
architecture".
For example - some proprietary SSD rackmounts
use significantly less memory chips and electrical power to deliver a usable
SSD capacity and resilience (compared to open systems) - due to the fact that
they have been optimally designed as a rackmount solution. (In many open SSD
arrays there is duplication of effort and wastage of performance due to
optimization at the module level followed by another level on optimization at
the array of modules level.)
The advantage of open SSD rackmounts
is they appeal to users who want to reduce the risks of buying from new
vendors - and who don't want to get locked into proprietary systems.
The argument goes something like this..
Even if the original
supplier of the proprietary
2.5" SSDs (used in
the open array) exits the market - there are another 30 or so manufacturers of
similar SSDs who can fill the unused / future disk slots at similar cost.
This
approach is likely to appeal to conservative buyers - and may carry more
weight than arguments about superior performance. Many customers will get "adequate"
performance from these "safe, dull and open" SSD architectures.
And even if they cost more today (than the proprietary SSD systems) - there's a
realistic expectation that in future they will cost less (or get faster) due to
competition in the SSD component / module market.
As you can see - the
rackmount SSD market is far from simple. And there will be no such thing as a
clear and present leader or winner which will satisfy all customers,
applications or budgets until another 2-3 SSD product generations have been
tested in the market. So it could remain a confusing picture till 2013. | |
| . |
more articles in this
enterprise series
Are MLC SSDs
Safe in Enterprise Apps?
Data Integrity
Challenges in flash SSD Design
RAM SSDs versus Flash
SSDs - which is Best?
Can you believe the
word "reliability" in a 2.5" SSD ad?
Rackmount SSD Trends -
open vs proprietary architectures
SSD ASAPs - Auto-tuning SSD
Accelerated Pools of storage
the Problem with
Selling Revolutionary SSDs to Risk Averse Users
Legacy vs New Dynasty
- the new way of looking at Enterprise SSDs |
| . | |
 |
| rackmount SSDs |
If he had his way... Sir
Squeaks-a-Bit
would stretch all 15K RPM disk pretenders
on the
rack and then remove their wobbly heads. | |
| . |
|
|
| . |
|
|
| . |
| this way to the Petabyte
SSD |
In 2016 there will be
just 3 types of
SSD in the datacenter.
One
of them doesn't exist yet - the bulk storage SSD.
It will replace the
last remaining strongholds of
hard drives in the
datacenter due to its unique combination of characteristics, low running costs
and operational advantages. |
 |
... |
The new model of the
datacenter - how we get from here to there - and the technical problems which
will need to be solved - are just some of the ideas explored in this
visionary article. | | | | |
