"With a 6:1 data reduction ratio, the effective capacity of a fully packed cluster is 1.9PB." ... why not claim even higher. It's like going to the bookies and saying I'm going to bet on the higher odds horse so I can win more. There's a reason the odds are higher! Anyone using 6:1 to arrive at their effective capacity is heading for a fall!
152 posts • joined 12 Aug 2009
From what I have heard about the way Autonomy recognized revenue, and forecasted, there were certainly some 'unusual' practices. And when I say heard, I mean been told first hand by current employees. I'm not sure what is legal and what isn't, or what HP should or shouldn't have been able to uncover prior to the acquisition but their practices were different to anything I'd seen before in a career of nearly a couple of decades.
Their win rate against other vendors? I think you can flip those numbers on their head vs HP!
In constant currency, the numbers were considerably better! Let's see what other companies post given the $ rate at the moment. Storage up something like 3% at constant currency, in a declining market, so dig a little deeper in the numbers!
Bearing in mind how average the XIV performance is when it's on very specific, tuned hardware, I can't imagine it working that well on any old server as SDS.
Quiet news day in The Register office?
Look, we're at the point where it's not a case of thinking 'why would I buy an AFA' but 'why wouldn't I'? If an AFA is cheaper per effective TB (based on a sensible low realistic de-dupe rate not some higher aspirational number) then why wouldn't you? Even if your apps don't need typical flash performance, if flash is cheaper then you'd buy it simply because it means you can put more on the array and not worry about it. And even apps that don't need performance as such aren't going to turn their nose up at it if it's available for less than the price of spinning disk. Forget flash being a luxury that few can afford, flash is cheaper than spinning disk for an awful lot of environments! It certainly is for the vendor I work for who has the best AFA on the market: http://searchsolidstatestorage.techtarget.com/feature/HP-3PAR-StoreServ-7450
We're getting beyond the point where customers are buying flash for pure performance these days so benchmarks and performance POCs are largely pointless. Flash is fast, we get it! But functionality whilst being bloody fast is the battleground now. It's very rarely a pure drag race these days, and for the small number of customers where absolute raw ridiculous performance is key, they already know where to look. There's not many workloads a few hundred 1000 IOPS at sub ms latency won't swallow up, but how easy is it to live with the array?
Re: Only EMC's View
What utter nonsense. Nobody buys an all flash array because it is an AFA, they buy it because it is bloody fast. HP 3PAR is bloody fast (and can be described as an AFA if you wanted to be pedantic). And the advantage is it has all the functionality that EMC seem to think no one can deliver!!
Any customer buying based on their claimed 6:1 is taking a massive gamble. If that's the average, across their entire fleet, then there is much more scope for high outliers, than low ones. And their arrays aren't very big in raw terms so buy an array needing 6:1 to get the numbers to work, only get a still fairly respectable 4:1 and then what? You don't have enough capacity and you're into a new box and going back cap in hand for more budget! Much better to buy based on a conservative dedupe ratio, and if you exceed it then it's free capacity!
The IBMer getting very defensive there! 23% margin in the sectors that IBM remains in is nothing to write home about! Wasn't their justification for selling off all the bits that have gone that they wanted to focus on higher margin areas? IBM have a number (a declining number) of customers welded to them, but I don't see them winning any new business anywhere! This downward spiral doesn't look like finishing any time soon.
No story here. This happens every year! And having moved from IBM, I love it as it used to be a real pain in the backside there ... but then again, 31st December is their year end. As for being forced to take time off, you are given 25 days of which HP can mandate up to 4! Hardly Draconian! And carry over is at BU discretion too.
Re: Avoid at all costs
I'm not surprised dedupe isn't 'working right' as it's not out in the field really yet and you sound so inexperienced (if you are a real customer) that I'd be amazed if you were on an early release programme!
No, not either ... both! You can have some SSD capacity boosting an AO Policy, you can have some volumes pinned in SSD and deduped, and you can have some SSD capacity used as Flash Cache. That's pretty flexible, in my book!
Re: ONTAP Performance
The F400 result was for a 4 year old, entry (and now dead) array running only spinning disk and the NetApp result from earlier this year couldn't even beat that despite having some flash! A disappointing outcome, however you try and frame it. I am genuinely interested to see how that claimed performance stands up when you actually start trying to do something useful with the array. People are not going to buy a NetApp array based on pure grunt so if they numbers disappear over a cliff when you turn on various functions then they are irrelevant. Also interested why the 1.3ms response time rather than the assumed 1ms typically used for flash numbers.
It has been a slouch ... their recent SPC-1 result (80k ish from memory) was about the same as a now retired 3PAR F400 (4+ years old). NetApp have never been about performance. I'd love to see if you can turn any functionality on and get anywhere near that 350k number.
I've never understood firstly why anyone still buys ds8000, but especially why you'd spend all that money on an enterprise array to dumb it down behind Svc!
You're helping them fix a problem that you created? That's good of you!
Re: Glass House
But you're not developing 7 mode any more so it's a transition that customers HAVE to make if they want to stay with NetApp. Yeah, it's not quite the same as the XtremIO but it's not that different either,.
This is a joke. NetApp has missed the boat so badly and to then come to market with such an under spec'd machine is laughable. What idiot is going to buy one of these first models? There are about 32 startups and about 5 big vendors with considerably more advanced offerings! It's surprising that NetApp have allowed themselves to get in this mess really. The end is nigh!
Re: look at their products and more specfically their customer support
Hodge Podge Portfolio? HP pretty much only sell 3PAR. There are a couple of other products around the edges but HP probably has the least 'hodge podge' portfolio of anyone. Even NetApp that used to have the cleanest portfolio has acquired and is developing net new products.
The bigger these claims of capacity savings become, the more the scope for a significant mis-sizing! As we all know, 'your mileage will vary' so the bigger that claim, the bigger the margin for error! Given the skepticism that even still exists in some corners around technologies like thin provisioning, I can imagine many IT shops having significant concerns over sizing, especially when close to the upper limit for these relatively small AFAs.
Why do we need to have a distinction for SSA vs arrays that can have HDDs installed in them. People don't buy a SSA because it is a SSA, they are looking for an array with certain performance characteristics and in the current market that is most likely to be met using SSDs but not necessarily just SSDs. Gartner talk about the ability to put HDD in the same array as SSD as if it is a bad thing. Surely flexibility can only be a good thing? If you can deliver the required performance without having to go for a completely separate silo of storage then why wouldn't you?
Re: Is there a comparison between the different flash vendor offerings?
Joking aside, with dedupe 3PAR now looks the real deal. One of the most scalable flash offerings, very very quick (certainly on a par with all but the most rapid of AFAs), fully functional, 6 x 9's, and flexible in that can have hybrid options as well. Very few companies will have just flash arrays (for some time) so if you are going to have a mixture of flash, hybrid and disk arrays, then 3PAR avoids the need for separate silos.
Re: why are you including Tegile
At $2/GB, you could argue that 3PAR is in the commodity space too. For a fully functional array, delivering 6 x 9's availability, to be in that price bracket is impressive!
Re: Is it yellow or orange?
Too keen with enter key... Offering 6 x 9's availability and a lower price per GB. Why go for a niche product when you don't have to any more.
Re: Is it yellow or orange?
This sounds more scalable and more functional than Pure so looks like the similarity ends at the pitch. HP do have their flash strategy bang on. Why have a flash silo when you can have a tier 1 array, offer
Re: Why AFAs are small
If you're spending money on an AFA then you do so for a reason and you want to thrash it ... so it would concern me that doing just that could cause problems!
And in terms of why they are so small, your second point is irrelevant as you're first point says that they can't currently scale due to technical limitations.
And as for your second comment, the AFA's are saying that flash is cheaper than disk so why should that only apply to smaller requirements? If it's cheaper, it's cheaper ... surely economies of scale only enhance that for bigger arrays? What you're also saying is that hybrid arrays are the way forward ... flash where it is needed, spinny stuff when it is needed.
So in defending Pure what you've actually said is that its' architecture won't scale, and flash is actually more expensive than disk despite their claims.
And whilst I'm at it ... what is an 'AFA customer'? There is no such thing. Find me any customer that doesn't have a variety of performance requirements across their environment. There is no such thing as an AFA customer. There are customers who have a performance requirement for some of their data sets but those same customers will have cold data sets so are they a 'spinny disk customer' too?
Pure switch off inline dedupe when the box gets busy. And why are all these afa's so small? I know they claim usable capacity as being after dedupe but even then they top out about 100tb. Why not just make a bigger one. Makes you think that there is some sort of scalability issue they are gradually working out.
Re: Thinking outloud $/GB coupled with Pooled vs. Persistent VDI Desktops
Have a look at a 3par 7x00. Afa now if that's what you want but flexible moving forward too. And well under the $/GB numbers talked about here.
Everyone has that problem though so don't try and make out IBM are suffering more than the rest. The fact is that customers are moving more to a 'good enough' approach and midrange arrays are way more capable than before. Everyone needs to sell a lot more than they used to to get to the same $ amount. Yet IBM seem to be doing a much worse job of that than most of the other vendors. Everyone is shifting more capacity than they used to so that defence is weak!
IBM storage is dead! The numbers are tanking and without a server biz, they have nothing to attach to (historically their main strength - very very few people have IBM storage without having IBM servers - I sold IBM storage for years and it is almost exclusively to 'house' accounts). The features that are mentioned in the article are them just trying to catch up with the competition and the absence of dedupe, and fairly mediocre performance mean that they are struggling to stand out! Dell and Netapp are waning too, HP 3PAR and EMC VNX2 are the only 'traditional' arrays that are still growing and the rest of the interest in the market is towards AFA's, hybrids and SDS.
The F400 is a 5 year old, now obsolete, array using no SSD or flash and still beats this latest greatest array from NetApp. Really? If you were NetApp, you wouldn't even publish this! And surprised how poor the V7000 is without SSD.
Couple of IBMers really feeling the need to get all defensive! So storage software like DS8000 array licenses, etc, are already rolled into the storage number so don't try to suggest that their storage numbers are underrepresented. IBM has had too fragmented a product line for too long. Yes, most vendors have multiple products, but there has been more overlap and confusion with the IBM portfolio than most others. The nSeries, DS5000, SVC/V7000, XIV and sometimes even the DS8000 have all been sold into the same space at times. IBM seems to have phases where one product is the answer, what is the question. It's been XIV but they seem quiet on that one now, with V7000 being the current trend. But regardless of how sound your logic is about a breakeven hardware business being worth being in because of the traction it provides for more profitable business lines, IBM isn't interested in hardware. I worked as a hardware salesman for IBM for over a decade and always felt that hardware was of no strategic importance to the company. Most of the xSeries guys have gone to Oracle, most of the storage guys have gone to HP, and important as Power is, it is a declining market too.
Poor analogies aside, the truth is that EMC is built on an aging and inflexible architecture and is struggling to keep up with the functionality and performance being offered by more modern architectures, of which 3PAR is one of many! And the point about the fork lift upgrade is not that moving to 3PAR isn't a forklift upgrade, but that if your upgrade is as complicated whether you stay with EMC or move to another vendor, then now could be a good time to move to another vendor.
I'm pretty sure most vendors (I work for HP, worked for IBM) will only provide free firmware updates for storage arrays as long as you have a support agreement in place. I didn't realise that HP ever provided firmware for unsupported arrays anyway. I don't think it is unreasonable to expect a valid support contract to be in place to receive enhancements to existing code.
Re: Now serving -- fast data with data reduction only.
Considering they claim 6:1 for average environments giving $3-6 per GB but as low as $2 per GB for VM then your 8.8:1 for VM puts their figures in doubt!!
It's not really saying anything that people don't already know ... StoreServ is the 'best' midrange array. And the reason is that it's an enterprise array masquerading as a midrange array, at midrange pricing. All the other arrays listed are classically midrange.
Even IBM don't seem to be pushing XIV that hard these days, they are all V7000 at the moment. It does change all the time though so maybe XIV will be flavour of the month again soon,.
Seriously Eric, are you for real? HP has a much simpler portfolio than EMC. EVA and VTL (I guess you mean VLS) have both already gone end of life so HP have a single dedupe solution in StoreOnce ... EMC have Data Domain and Avamar. P4500 is an iSCSI only solution. In terms of block storage, if it's very low cost it's MSA, if it's 6 x 9's it's P9500, but everything else (95% of use cases) is 3PAR. So from 8 drives, to nearly 2000 drives, from SSD to Nearline, it's all 3PAR. EMC has VMAX, VNX and XtremeIO all covering that space and no interoperability.
HP and NetApp have a much more similar model. Buy into the benefits of 3PAR / NetApp and then just pick your size. You don't need to worry if you're small or huge, you have access to all the same functionality.
Re: Gave up tape and removable storage years ago
How many years have we been hearing that 'tape is dead' yet it's still going relatively strong. All these newer, faster tiers don't replace anything, they just give more choice and flexibility. There will be storage scenarios that need tape, there will be storage scenarios that need decent performance and decent capacity, and there will be storage requirements that just need blistering fast. And most people have all 3, and more. Flash doesn't replace disk or tape as such, it just means that for apps that might have needed huge numbers of spindles and wasted a shed load of capacity as a result, there is now a better way! The clever vendors are the ones who will offer all these choices as seamlessly as possible. The HP approach of having one enterprise featured array that can offer pure flash, hybrid, and SAS is a good one. The same addressable market for EMC is covered by about 5 different arrays! Same for most vendors, to be honest.
Re: Sales dropping the ball at HP?
Frank, I've tried finding you on LinkedIn but to no avail. If you are still in the market for the capacity then maybe the admin of the site could exchange our details? I do agree with the other poster who replied saying that putting 3PAR behind SVC is diluting the benefits of 3PAR, but the 7000 range is very cost effective so you could still see a benefit in going with 3PAR. The new MSA's will absolutely fly so they could be an option if supported.
So the quote says they have most of their success against NetApp, EMC and Dell, yet HP are the ones you mention in the title? Axe to grind?
Re: Storm clouds
Rewards how? It would mean they have to buy in the hardware components for their XIV, V7000 and probably some other storage platforms, or develop proprietary platforms for them which increases cost and slows down development. And also a hell of a lot of commodity servers still get bought, and it's a lot easier to attach storage to a server sale than it is to compete with someone who is!
Re: Storm clouds
Although the x86 market isn't an overly profitable one at the moment, there are other benefits for the likes of Dell and HP in remaining in it, less for for IBM which has already ditched personal systems. The buying power and economies of scale help prop up other platforms, like personal systems and storage. Many of the storage solutions on the marketplace are increasingly based on x86 server platforms, and maintaining control of design and manufacture of that is useful. Also, there is a lot to be said for the market presence and penetration that servers get you. I'd wager that HP and Dell sell a hell of a lot of other 'stuff' on the back of commodity server sales. I'm not being critical of IBM's strategy as I think it is also a valid one - the less valid approach is a half way approach where all benefits from either strategy are diluted so for IBM to increasingly get out of commodity markets is another move in the right direction.
As you correctly mention at the start of your post, you have no knowledge or understanding of 3PAR. 3PAR doesn't use traditional RAID sets either, widestriping all day across all drives in the array, so the management issue you raise isn't really a point. It also allows many 3PAR customers to drive extremely high levels of utlisation, greater than 90% in some cases. Compare that to the huge amount of wasted space in the XIV! As I said, a 243TB 3PAR system using higher performance drives and with wide striping so no hot spots, uses 30% less space than an XIV. The arguments you raise against VMAX are the same ones 3PAR would raise, so you are validating the 3PAR messaging. The difference is 3PAR is a lot more efficient both in terms of space and power.
XIV is definitely an 'interesting' architecture for the 'cloud' but has 2 main issues - power and density.
243TB usable in a rack really isn't that special and when you're building a multi PB cloud then density matters. And even more important is power draw. XIV3 used 8.4kW ... that's a lot of power for 243TB. A comparable 3PAR array, even if you feel the need to spec it fully with SAS drives rather than 7.2k nearline, only takes 31U and 5.4kW. And there are much better ways of hitting 243TB, with good performance, so in reality a 3PAR array would take less than half the space and less than half the power. Could be why 3PAR is in 8 out of the top 10 service providers.
XIV isn't really dense enough for big data, or have enough performance for Tier 1. It does seem a reasonable fit for that Tier 2 type space but so do a lot of other products!
Re: A bit of a waste for Violin to go to Dell
Is there a lot of IP in Violin? Don't they just use Symantec for the functionality? I think the hockey stick growth opportunity for Violin has been and gone ... this market is getting more competitive daily and there are players appearing with much bigger teeth than Violin. When EMC, IBM and HP really get their act together them the smaller players will struggle.
There will be probably be Intel versions of the cartridges as well.
HP sold Violin arrays in theory only, the relationship never got off the ground.
In terms of latency, if you want to share Violin arrays across a few servers then you have to introduce a gateway (x86) server which introduces more latency, and then you still have the fabric latency as well. That's a worse solution than having a genuine storage array that can delivery ridiculous performance and share that as standard. If you just have one server wanting blistering performance then maybe I can see your point, but to be honest, there are even better and lower latency ways of achieving that than Violin.
And the bottleneck in traditional arrays with SSD is the controller not the network. VNX, EVA, etc, all flood the controllers after only a very small number of SSDs. 3PAR can support 200 SSDs without that issue. You're picking at straws. The 3PAR solution might not be absolutely as fast as a little flash only box but is still 500k IOPS+ so plenty for all but the most demanding of apps. But the 3PAR delivers that performance without sacrificing usability.
See, I thought this article was mostly about XtremeIO. Each brick apparently contains 16 x 200GB SSD's but now you're telling me that this isn't a flash array. Sounds pretty flashy to me, as does a StoreServ 7400 with 200 x 200GB SSDs.
And there is a difference between throwing a few SSDs in a disk array and having an architecture that lends itself to supporting very large numbers of SSDs! Throwing a few SSDs in an array will not delivery hundreds of thousands of IOPS at sub millisecond latency ... StoreServ will. Just because HP don't need to bring to market a new specific flash array, doesn't mean they don't have one already!