Low end SAN for ESX and Hyper-V

[Thuleman]

Exactly.
The PS series is actually really nice when you have multiple units, the software makes replication setup, options, and scheduling a breeze. That EQL just bundles the software with the unit is imho the major selling factor. There are no hidden costs, no software license "gotcha!" moments.

I only had to call support once because I needed to reset the unit and that works though a challenge-response method that needs to be called in. The wait time was under 5 mins, there was North American support, and the issue was resolved quickly.

I am sure from a business perspective the EQL units are a good value, I am in academia though and while 23K isn't that much for us either, it does every now and then irk me that I did't just white-box it and used the extra cash to buy more toys.

Still the PS5000E is a solid foundation, and my expansion will either be pulling the Dell disks out and sticking in my own 1TB disks one of these days, or I will buy a white-box solution when there's a need for more fast SAN storage.

 

[AreEss]

What is it about the EqualLogic arrays that you do not like? Everything that I have heard (from people that are actually using them in production) and read about the arrays has been positive. I am by no means trying to defend EqualLogic. I have no dog in this race. I am just curious as to why you feel this way. Thanks for any info you can provide.

The price-performance is obliterated by everyone else in the market. Badly. Their RAID options are beyond limiting, to the point where the array was quickly found to be completely useless for serious use. Just because people don't claim to have performance problems, doesn't mean it has any scalability - it doesn't. In order to scale these, you have to fork over for multiple, multiple, multiple heads. This creates a management nightmare, to put it mildly. For the same dollars, I can deploy a scalable FC solution based on the IBM DS3400 using SAS disks. Or a scalable LeftHand from HP. Or any other number of other options.

They also lied through their teeth on numerous occasions. They claimed failover and redundancy - nope. They claimed that it was faster than FC with 2 GigE connections - that's just flatly absurd and technically impossible. They claimed to be cheapest - we were quoted over $70K for a single unit with SAS disks, which could only meet half the capacity requirement.

The fact is that Equalogic hardware is neither scalable, nor good value for the money, in ANY business case. They are flat bricks. Any business that intends to grow can obtain much better value for their dollars with other solutions from IBM, Dell, and HP. And any business that claims they don't want to grow, is lying through their teeth. Rebuying the same controller over and over to expand your disk is not how anyone else works, for a reason. You buy disks and enclosures to expand disk, not controllers.

Let me spell it out for the pro-Equallogic camp: YOU ARE WRONG. EQUALLOGIC IS POOR VALUE, AND DOES NOT FIT THE REQUIREMENTS. Go argue with somebody who can't read.
To meet his storage requirements, he needs a minimum three PS6000E's in a group (presuming he doesn't want to run RAID0,) which requires two additional managed ethernet switches, creates more management headaches, and requires special network and firmware settings to perform failover (which is known to be iffy.) And 16x1TB for it is going to cost just as much as 16x1TB from EMC or IBM or HP.
In numbers? EqualLogic PS6000E for 20TB will cost me ~$30K x 3, plus switches, so call it $105K. Because I have to buy identical PS6000E's. IBM DS3400? $23,863 for the first 12 1TB disks (we'll say RAID6, so 8TB usuable.) Then to expand it? A shelf is $3,200 and each 1TB disk is $1,200. Which gives me 24TB at a price point of about $60K list. Figure your typical discount from IBM is 30% on these, which drops it to about $42K or less than half the cost of the EqualLogic even when you add in FC switches. (Presuming Qlogic SANBox 1400's which are $2.5K each, so another $5K.)

 

[AreEss]

Still the PS5000E is a solid foundation, and my expansion will either be pulling the Dell disks out and sticking in my own 1TB disks one of these days, or I will buy a white-box solution when there's a need for more fast SAN storage.

Like ALL enterprise and psuedo-enterprise arrays, you cannot just swap in your own disks. One, there are firmware locks on the EqualLogics. Two, it immediately terminates all support agreements.

 

[mct]

The price-performance is obliterated by everyone else in the market. Badly. Their RAID options are beyond limiting, to the point where the array was quickly found to be completely useless for serious use. Just because people don't claim to have performance problems, doesn't mean it has any scalability - it doesn't. In order to scale these, you have to fork over for multiple, multiple, multiple heads. This creates a management nightmare, to put it mildly. For the same dollars, I can deploy a scalable FC solution based on the IBM DS3400 using SAS disks. Or a scalable LeftHand from HP. Or any other number of other options.

They also lied through their teeth on numerous occasions. They claimed failover and redundancy - nope. They claimed that it was faster than FC with 2 GigE connections - that's just flatly absurd and technically impossible. They claimed to be cheapest - we were quoted over $70K for a single unit with SAS disks, which could only meet half the capacity requirement.

The fact is that Equalogic hardware is neither scalable, nor good value for the money, in ANY business case. They are flat bricks. Any business that intends to grow can obtain much better value for their dollars with other solutions from IBM, Dell, and HP. And any business that claims they don't want to grow, is lying through their teeth. Rebuying the same controller over and over to expand your disk is not how anyone else works, for a reason. You buy disks and enclosures to expand disk, not controllers.

Let me spell it out for the pro-Equallogic camp: YOU ARE WRONG. EQUALLOGIC IS POOR VALUE, AND DOES NOT FIT THE REQUIREMENTS. Go argue with somebody who can't read.
To meet his storage requirements, he needs a minimum two PS6000E's in a group, which requires two additional managed ethernet switches, creates more management headaches, and requires special network and firmware settings to perform failover (which is known to be iffy.) And 16x1TB for it is going to cost just as much as 16x1TB from EMC or IBM or HP.

I'm not saying that EqualLogic may be a good solution for him now based on what he needs. However I can tell you that I can get two PS6000XV filled with 450GB 15k SAS drives for a total of $70k. From all of the reading and research I have done I have never seen any independent review bash the product like you have. In fact everything I have read thought it was a good product. If you can provide some links I would really like to read about what you have said. I have no doubt that they lie about what their product can do. I have never met a salesman that didn't lie about their product including the ones from NetApp and EMC.

 

[mct]

Exactly.
The PS series is actually really nice when you have multiple units, the software makes replication setup, options, and scheduling a breeze. That EQL just bundles the software with the unit is imho the major selling factor. There are no hidden costs, no software license "gotcha!" moments.

I only had to call support once because I needed to reset the unit and that works though a challenge-response method that needs to be called in. The wait time was under 5 mins, there was North American support, and the issue was resolved quickly.

I am sure from a business perspective the EQL units are a good value, I am in academia though and while 23K isn't that much for us either, it does every now and then irk me that I did't just white-box it and used the extra cash to buy more toys.

Still the PS5000E is a solid foundation, and my expansion will either be pulling the Dell disks out and sticking in my own 1TB disks one of these days, or I will buy a white-box solution when there's a need for more fast SAN storage.

I am in academia as well and the people that I have talked to about EQL are as well. Where I work we currently do not have a SAN in place and this is something that I have been researching for about a year now. I will say that I have been looking really close at NetApp and EMC as I think they are the best in the business at this point. I am just suprised about some responses in regards to EQL when I have heard nothing but the exact opposite from others.

 

[AreEss]

I'm not saying that EqualLogic may be a good solution for him now based on what he needs. However I can tell you that I can get two PS6000XV filled with 450GB 15k SAS drives for a total of $70k. From all of the reading and research I have done I have never seen any independent review bash the product like you have. In fact everything I have read thought it was a good product. If you can provide some links I would really like to read about what you have said. I have no doubt that they lie about what their product can do. I have never met a salesman that didn't lie about their product including the ones from NetApp and EMC.

And you probably believe Storage Anarchist too. In the Enterprise arena? There is no such thing as an independent or unbiased reviewer. Everybody has their favorites, everybody gets played by sales, and the list of things goes on. Ethics and equality are completely nonexistent in the arena, and anything bad gets hidden behind NDAs. I'm still prohibited from publishing 99% of my data because of NDAs that were signed by management or were required just to get a sales presentation. (Something IBM is incredibly grateful for, after one technical presentation in particular.) That way if and when something sucks, you can't say anything about it. There is information I have on the UltraSPARC RK (Rock) that I can never publish because of NDAs.
Most days, I really do think I'm the only person in the whole field who has absolutely no loyalty to any vendor, despite what others repeatedly claim. Sorry folks, fact is, I'd sell out to Sun tomorrow if they had a superior product. Or HP. Or EMC, if they gave me better value. I'm a money, function and reliability guy. If Joe's PC can give me the same functionality and reliability for less money, I'd sing his cost-performance praises from rooftops.

 

[mct]

And you probably believe Storage Anarchist too. In the Enterprise arena? There is no such thing as an independent or unbiased reviewer. Everybody has their favorites, everybody gets played by sales, and the list of things goes on. Ethics and equality are completely nonexistent in the arena, and anything bad gets hidden behind NDAs. I'm still prohibited from publishing 99% of my data because of NDAs that were signed by management or were required just to get a sales presentation. That way if and when something sucks, you can't say anything about it. There is information I have on the UltraSPARC RK (Rock) that I can never publish because of NDAs.
Most days, I really do think I'm the only person in the whole field who has absolutely no loyalty to any vendor, despite what others repeatedly claim. Sorry folks, fact is, I'd sell out to Sun tomorrow if they had a superior product. Or HP. Or EMC, if they gave me better value. I'm a money, function and reliability guy. If Joe's PC can give me the same functionality and reliability for less money, I'd sing his cost-performance praises from rooftops.

I am not here to dispute your claims. This is something that will affect me eventually. I really want unbiased opinions if I can get them so any info that you can provide would really be appreciated. I am not sure if I am coming across well in my posting however there really isn't a need to get a hostile attitude with me. I am just a guy looking for information. This is something that you seem very passionate about and I hope that you can send me in the right direction for this information. By the way I have no idea who Storage Anarchist is.

 

[Thuleman]

AreEss, it's clear that you have a problem with EQL for whatever your reasons are. That's perfectly fine as long as you don't claim the ultimate truth and state that everyone else who doesn't subscribe to it is an idiot or illiterate.

EQL is in business, EQL is on the very top of iSCSI, whether you like them or not doesn't change the fact that storage professionals world-wide are buying EQL arrays ..., but wait, they are all idiots after all, of course ...

The bottom line is that especially for academia the EQL is a great solution exactly because academia is most often NOT an Enterprise environment. It's easy to spend money this fiscal year, but it's also easily impossible to come up with cash to support the same system next fiscal year. Storage arrays which require yearly licensing of the software to actually administer them are out.

YES, there is a price step when going from one to two units, BUT you can easily trade in your old one for a new and bigger one (+$) and then take delivery of your new one and effortlessly migrate your old SAN to the new SAN without to incur additional licensing costs for migration tools (unlike other "unnamed" vendors). In addition the data growth in academia is much more foreseeable and happens at a far slower pace. Scalability (or the price at which it is achieved) isn't a major factor. You simply can not beat EQL in the iSCSI market for price/performance of a single unit. Take my 23K unit and prove me wrong, since you do seem to know oh-so-much about it.

BTW, the thin provisioning the EQL provides is awesome in an ESX environment.

I am all for using the right tool for the job, but let's make sure that:
- the requirements are actually true
- the requirements are understood
- no assumptions about the customer's business model are made
- the right tool is suggested for the job

If we do that, then EQL is in the race. It may not come out ahead, but it's definitely a contender.

 

[mct]

The bottom line is that especially for academia the EQL is a great solution exactly because academia is most often NOT an Enterprise environment. It's easy to spend money this fiscal year, but it's also easily impossible to come up with cash to support the same system next fiscal year. Storage arrays which require yearly licensing of the software to actually administer them are out.

This is exactly our problem and the problem of others. From a support cost standpoint EQL is by far the cheapest especially when compared to NetApp and EMC. It is also very hard for us to come up with $20k just to purchase additional software and licenses that those like NetApp come out with. With the EQL solution you get all of the software available. There is no nickel and diming you to death. With that said I still think NetApp and EMC make great products.

 

[AreEss]

AreEss, it's clear that you have a problem with EQL for whatever your reasons are. That's perfectly fine as long as you don't claim the ultimate truth and state that everyone else who doesn't subscribe to it is an idiot or illiterate.

EQL is in business, EQL is on the very top of iSCSI, whether you like them or not doesn't change the fact that storage professionals world-wide are buying EQL arrays ..., but wait, they are all idiots after all, of course ...

Hi Shill. Or Zealot. Take your pick.
Nobody on this earth who isn't in EQL's back pocket one way or another claims they're "on top" of iSCSI. Everyone has long acknowledged that Adaptec is the absolute bottom, and right now NetApp sits on top. Nobody competent is going to honestly claim, or legitimately prove, that a product with no expansion beyond re-buy and that does "expansion" by building logical arrays over arrays over Ethernet host paths is "top of the line." There are technical reasons you don't do these things. Not to mention that the 16TB raw maximum capacity of a single PS5k and 9.4TB raw maximum capacity of the PS6000S group puts them at the absolute bottom of the capacity and efficiency ladder.
Oh boy, a "user friendly" (i.e. any mook with a browser who can do 2+2=4) interface. News flash; the UI does not a product make. Even the Dell/EMC AX4's are more competitive.

I'm sorry if you don't like technical facts. That doesn't change that they're facts.

 

[Thuleman]

Nobody on this earth who isn't in EQL's back pocket one way or another claims they're "on top" of iSCSI.

I'm sorry if you don't like technical facts. That doesn't change that they're facts.

... I am glad we can agree on facts anyway. How about sales stats?

IDC: Recession Driving Down Storage Spending

Dell led the iSCSI market in the quarter with more than 36 percent of revenue, followed by EMC with just under 16 percent, IDC said.

EQL = iSCSI, there's no point in considering FC or NAS or whatever, apples to apples Dell leads the iSCSI market. Feel free to dig through the quarterly reports yourself to find the sales numbers.

The bottom line is that junk doesn't continuously dominate the sales stats. Hype does last, but users do eventually catch on, and EQL has been around long enough for that to have happened already if it were true.

From the same article:

Sales revenue from entry-level storage systems, priced below $15,000 per system, grew 9.9 percent. Much of this came from large enterprises buying these systems to meet immediate capacity needs rather than investing in the larger platforms they will eventually need

Which proves that people make the choice to buy what they can afford, rather than what they actually need, which is very much the subject of this thread.

 

[AreEss]

... I am glad we can agree on facts anyway. How about sales stats?

EQL = iSCSI, there's no point in considering FC or NAS or whatever, apples to apples Dell leads the iSCSI market. Feel free to dig through the quarterly reports yourself to find the sales numbers.

Except that Dell is comprised of Dell, EqualLogic, and Dell-branded EMC. Oh dear. Did I just blow up your numbers? Seems I did.

The bottom line is that junk doesn't continuously dominate the sales stats.

Yet McDonald's sells billions of cardboardburgers, and Dell sold millions of desktops with the most miserable QA and support record.
Junk does continuously dominate sales stats. Because it's cheap junk, and many businesses are stupid enough to believe the idea that "oh well it was cheap," while they're burning time, people and more money than it would have cost to do it right, trying to keep that cheap crap working. Don't bother insulting people's intelligence by claiming that's not true - anybody here who actually WORKS in any shop knows that's a fact. When was the last time a business insisted on the more expensive copy paper that didn't jam instead of the extremely cheap stuff that jams so often it's wrecking copiers? Uhhuh.

 

[Thuleman]

I think we are in a circular argument at this point. You say EQL is junk, I say that it isn't.
You base your opinions on theorycraft and a review unit from years ago, I base my opinion on actually running one in production. Readers will form their own opinion.

The PS5000E works for me in the environment I am using it in, it was the least expensive among highly overpriced solutions, and it does what I need it to do (as a whitebox solution would have).

At the price I paid no other vendor could compete. I set the array up in January, upgraded the firmware somewhere since then, and had exactly zero issues with it. Try to convince me I should have bought that 30k or 40k competing vendor unit and watch me giggle.

The reality of things is that EQL fills the niche where users need a SAN but could care less about the Amazon.com, craigslist, or [H]forum's of this world. Not everyone needs a million IOPS at FC speeds, and not everyone who doesn't need it is willing to pay for it anyway.

 

[sabregen]

The old "no one ever gets fired for buying Cisco"

No argument with your argument here, but the quote is, was, and (for the foreseeable future) will be:

"No one ever got fired for buying IBM."

It is a testament to branding, and this quote has been circulating since at least the early 1980s. Here's a book on branding, where it's mentioned:

http://books.google.com/books?id=Yn...2qn7CA&sa=X&oi=book_result&ct=result&resnum=4

No, I'm not trying to be a quote Nazi, just throwing in my $.02

 

[AreEss]

No argument with your argument here, but the quote is, was, and (for the foreseeable future) will be:

"No one ever got fired for buying IBM."

Well, unless they updated to the 7.x firmware at release. Then they probably got fired. Right after the second data loss event in a week.

 

[lopoetve]

I explained it a bit poorly, probably.
ESX has the same problem, as does VIOS and everything else. Because you have an obfuscation layer in there, your guest has to wait for your host to run out of retries in any media error. In the meantime, it's either completely dead or stacking up more IOs. This is especially bad when virtualizing tape, for example, because a media error can take several seconds or more to pass up to the guest - which has been stacking up more IO the whole time it's waiting.

I can guarantee you ESX does not have that problem if you've configured right . The guest is seeing the same thing that the host does (at least, if you made the appropriate registry setting changes), and the host gives up well before the guest does and sends a "failed IO" message up. Yes, commands have been queued up like anything else in the scsi queue, but they'll fail/retry/failover just like physical, if you've set the IO timeouts appropriately for the clustering.

The MSCS in ESX guide tells you to increase the timeout setting so that the guest doesn't fail / succeed before ESX does - and ESX doesn't confirm the write to the VM till the physical layer does. This was done intentionally to guarantee data integrity on MSCS.

 

[lopoetve]

I am in academia as well and the people that I have talked to about EQL are as well. Where I work we currently do not have a SAN in place and this is something that I have been researching for about a year now. I will say that I have been looking really close at NetApp and EMC as I think they are the best in the business at this point. I am just suprised about some responses in regards to EQL when I have heard nothing but the exact opposite from others.

I'm going to have to quietly agree with you.

I've worked with many EQL shops and they've never had even the slightest bit of problem with them, performance has been solid for the price, failover works exactly as advertised (and SP failover works great - they were the first to get that going on iSCSI officially for ESX and certify the configuration), and the arrays are very easy to work with.

I've never had an issue with them, and most customers I've had haven't either. They're not NetApp or EMC, but they also cost a good deal less.

 

[lopoetve]

AreEss, it's clear that you have a problem with EQL for whatever your reasons are. That's perfectly fine as long as you don't claim the ultimate truth and state that everyone else who doesn't subscribe to it is an idiot or illiterate.

EQL is in business, EQL is on the very top of iSCSI, whether you like them or not doesn't change the fact that storage professionals world-wide are buying EQL arrays ..., but wait, they are all idiots after all, of course ...

The bottom line is that especially for academia the EQL is a great solution exactly because academia is most often NOT an Enterprise environment. It's easy to spend money this fiscal year, but it's also easily impossible to come up with cash to support the same system next fiscal year. Storage arrays which require yearly licensing of the software to actually administer them are out.

YES, there is a price step when going from one to two units, BUT you can easily trade in your old one for a new and bigger one (+$) and then take delivery of your new one and effortlessly migrate your old SAN to the new SAN without to incur additional licensing costs for migration tools (unlike other "unnamed" vendors). In addition the data growth in academia is much more foreseeable and happens at a far slower pace. Scalability (or the price at which it is achieved) isn't a major factor. You simply can not beat EQL in the iSCSI market for price/performance of a single unit. Take my 23K unit and prove me wrong, since you do seem to know oh-so-much about it.

BTW, the thin provisioning the EQL provides is awesome in an ESX environment.

I am all for using the right tool for the job, but let's make sure that:
- the requirements are actually true
- the requirements are understood
- no assumptions about the customer's business model are made
- the right tool is suggested for the job

If we do that, then EQL is in the race. It may not come out ahead, but it's definitely a contender.

Well, I'd put NetApp at the top of iSCSI (and NFS) personally, but EQL ain't bad in my experience. Especially given how good their support has been, in my experience. Those guys are really friendly and nice

 

[irrision]

Ahh the old DSxxxx is better/faster/cheaper than everything else debate. But really there is fairly close to price parity between manufacturers and if there isn't if you go to the vendor you want with a quote for a competing product model they will almost always match it or beat it. Right now is a great time to buy hardware, vendors are hurting and are letting things go damn near cost if you push them a bit. We've seen IBM, Netapp, and EMC come down to less than half their list prices on quotes this year vs the typical 30% discount you typically see in a good market. The main thing you have to watch out for our your ongoing maintenance costs when you are cross quoting as those are what really set vendors apart in most cases.

I'd also like to mention that looking at a FC attach DS series is not going to save you money over buying anything iSCSI based even if you do have to buy a good gig-e switch to plug everything into. After you spend on a fiber switch, SFPs, licensing, fiber, and HBAs you are going to burn up your budget fairly quickly.

On Equal Logic, you can get instant quotes online from a couple places if you want to get a feel for what they really cost.

Have you looked at some of the smaller vendors (other than Equal Logic?). I would take a look at things like Data Pillar, Xiotech, and Compellent. They have every reason to be more price competitive as they can't trade on their brand reputation yet and love to get references if you are willing to give them one (assuming you don't have a bad experience... obviously).