HP Blades vs IBM Blades Forum

This forum is to discuss Hardware comparisons including but not limited to blades, servers, appliances, and storages

HP Blades vs IBM Blades Forum

Postby forumadmin » Wed Apr 02, 2008 4:13 pm

Hi Everyone,

We know there has always been a big debate between IBM & HP Blades since both have announced their first blades. The debate and competition have only been extra heated with each newer release of both vendors. We have put quite good comparison of IBM BladeCenter vs HP C-Class chassis at IBM BladeCenter vs HP C-Class chassis (IBM Blades vs HP Blades).

Although we had put best of our efforts to be fair to both and cover as many points as possible, we would like to hear your opinion, experience, and debates of these blades. Please point any features we have dropped in our comparison and you believe its worth mentioning.

We are looking forward to hear your opinion.

Thanks to all,
ITComparison.com Team.
forumadmin
 
Posts: 36
Joined: Tue Dec 18, 2007 5:52 pm

Postby chrislynch » Wed Jun 18, 2008 9:35 pm

Your blog post has too many issues with it, I don’t know where to start. You call it an “unbiased” opinion, but I have to seriously object to it. I have to ask, did you even test anything you stated here, or did you simply plagiarize from the poorly written Clabby Antics IBM paid and published report?

So, I will pick apart each “section” you outline in your blog.
IBM has several chassis sizes with different specs and sizes to meet the special requirement of every organization and options which can be interoperated between different chassis, where HP is only offer the c-class chassis and their older p-class chassis without any interopobility between them at all.

While this is true that the current shipping solution from HP is the c-Class, it was for a good reason. There were too many limitations to the p-Class solution offering that did not justify keeping the platform for customers.

When comparing IBM’s solutions and looking at their “backwards compatibility” statements, one has to seriously wonder “what the heck are they smoking?” Can I support the more powerful HS21XM in the same original BladeCenter chassis? Not with higher than 80W QC CPU’s. Can I inter-mix their compact form-factor (CFF) cards or High Speed Form Factor (HSFF) with legacy blades (i.e. HS20, LS20) in the new BladeCenter H chassis? No. How about that new fancy QS21 and the original BladeCenter chassis? Nope. Ask yourself this: Why does IBM continue to post power specs for their original BladeCenter chassis and not their newer BladeCenter H chassis?
ITComparison Team Comments IBM are offering more blades platforms than HP specially with them including Power and Cell BE blades while HP are not offering any equivalent to these platforms, which can be a great advantage to customers who care to run operating system and applications which best supported on these platforms (Ex: AIX and Linux for power).

Other than Power, have you seen any real customers use or adopt Cel? No. Can it offer an edge over the traditional x86 platform for HPC? Only specialized applications, with a long and challenging development cycle could they prevail.

Plus, this argument is just silly. HP supports Itanium. IBM does not. IBM support Power. HP does not.

Again, can you install the new JS22 (Power6 offering) or the Cel-based blade into the original BladeCenter chassis? Here’s a hint: Nope.
ITComparison Team Comments It seems HP has succeeded to fit more blades per single chassis 16 blade vs 14 for IBM. As well HP can fit 16 blades with hotswap HDD where IBM can only fit 7 of them as IBM Blade requires an expansion unit to fit hotswap HDD, but when it come to redundancy IBM has a long wining of the race. IBM chassis can fit 14 redundant blades where HP only can fit 8 semi-redundant blades in their chassis. We called HP semi-redundant blades as they are not fully redundant as explained under the redundancy comparison.
Redundancy
- Dual power connections to each blade
- Dual I/O connections to each blade
- Dual paths through the backplane to I/O, power and KVM
- Single power connections to each blade
- Single I/O connections on BL460c and BL465c blades
- Single I/O paths for mezzanine slots 2 and 3 on the BL480c and BL685c
ITComparison Team Comments It seems IBM is a clear winner on blades redundancy at the moment. This can be a major decision factor for large enterprises, as it can be a major availability factor.

What does the term redundancy mean? More than two. Does IBM offer redundant traces for *EVERY SINGLE PORT*? No they do not. Just because they have two connectors to their backplane does not make it a redundant solution. Ask yourself this. Why did IBM stick with their same backplane design of combining active components (they have over 60 of them on their backplane) and power on the same PCB? If there is even one short the entire backplane is lost; So much for redundant connections.

HP offers separate signaling and power midplane so one does not affect the other. There are guide pins to make sure that someone does not install the blade incorrectly, and potentially introducing a short. Plus, there are ZERO active components on the mid-plane that would affect the operation of the chassis. Well, there is one. It only contains the FRU data of the chassis, and if it’s lost, then you lose the serial number and part number of the chassis.
ITComparison Team Comments It seems HP are having an advantage in being able to fit larger number of blades which includes HotSwap HDD into their C-class chassis, but IBM has a valid argument as most blades customers depend on boot from SAN which provide them with stateless blades and all kind of advantages including the ability of taking snap shots of their blades. In addition, with IBM introducing Solid State Drives it has even reduced the need for hotswap harddisks even further as these have no spinner and their reliability are way better than SCSI HDD. It seems HP still not offering Solid State Driver at the moment, but might be in the future.
It seems a decision of more blades with hotswap HDD per chassis versus a real redundancy is the greatest comparing factors between HP and IBM blades. HP can fit more blades with hotswap HDD where only IBM can offer a fully redundant blade and Chassis.

BfS is not trivial. It does take planning and support from a number of vendors. While BfS can offer a significant advantage, not all organizations have implemented BfS nor have the means to support BfS. This is yet another example of IBM try to find a way around their self imposed limitations.
- Light Path Diagnostics uses battery to help diagnose even without power to the blade.
- HP offer diagnostics LEDs beside some components, but will not led without power.
ITComparison Team Comments Better and faster serviceability in the IBM Blades with the ability to pinpoint the problem even if the blades is not powering up, which is not offered by HP.

This statement is a complete oversight of the Onboard Administrator and the Insight Display. Can IBM pinpoint to you what exactly is wrong? No. They simply tell you the area that is preventing the system from booting or where the issue is at. Not how to resolve it. Plus, go try looking up their boot sequence codes. They are not consistent across every system, and offer very little in the way of resolution.

With the HP OA and Insight Display, an admin can find out exactly why their blade is not powering on. Oh, you installed a mezzanine card into the wrong slot? We show you that, and tell you what you should do to resolve the issue. Not just some LED flashing.
Event identification
- First Failure Data Capture
- Nothing Equivalent
ITComparison Team Comments IBM Blades got a better non over-lapping error reporting through their Management Module which help in resolving cascaded problems faster.

HUH??? I cannot even decipher what you are trying to say here?
Integrated 4X InfiniBand® switch modules
ITComparison Team Comments IBM offer easier deployment and management of their InfiniBand switches as they are managed through the management modules where the ones offered by HP are unmanaged.

And not all customers will limit themselves to a single chassis with IB. They will connect the chassis to their exsiting IB fabric, which would provide the management tools. Of, if they choose not to, they can still use the Voltaire GridVision BladeFM and the Cisco High Performance Subnet Manager (HSM) as a host-based subnet manager. No need for external IB switches with embedded management.
Blade deployment and redeployment

- Open Fabric Manager, Uses standard switches, single login across 100 chassis
- Virtual Connect, Uses proprietary switches, single login across four chassis
ITComparison Team Comments
IBM Open Fabric Manager Feature HP Virtual Connect
All Ethernet and Fibre Channel switches—Cisco, Nortel, Brocade, QLogic Switch support Single proprietary HP Ethernet switch, Single proprietary HP Fibre Channel switch
Automated and integrated with resource pooling Failover support Requires manual intervention
Virtually all BladeCenter chassis, blades Compatibility Single c-Class chassis support
Single login via Advanced Management Module across 100 chassis Interface and capacity Separate login to Virtual Connect Manager across four chassis

The whole goal with VC is to reduce complexity and management in a customer’s environment, not just to simply provide “virtual hardware addresses” inside the chassis. With VC, the Network folks manage the infrastructure, and provide connectivity to the chassis. Outside of configuring network definitions (which is extremely similar to VMware’s vSwitch technology), they do not manage the embedded devices. Same with the SAN folks. In fact, they have even less to manage with VC.

With BOFM, you still either need to have pass-through (which does not resolve the too many cables issue) or switches (I have n more devices to manage on my fabric.) Oh, and “automated and integrated with resource pooling” is a feature of IBM Director and does not provide an automated process in moving a blade and it’s configuration from one slot to another. You still need to manage your chassis with Excel or your favorite worksheet editor to manipulate the CSV configuration. TO move a blade with BOFM, it’s Cut and then Paste. Yes, IBM does provide a GUI tool, but only with the BOFM Advanced version. But, it does not remove the necessity of the CSV file.

With VC, you have access to a CLI and a WebUI for management. You want to move a profile and a blade to another bay or to a bay in another chassis? Sure. It is quite easy with two easy mouse clicks or a single CLI statement.

Oh, can you mix QLogic and Emluex HBA’s inside the same IBM chassis with BOFM? No. Why? Because they use QLogic’s or Emluex’s WWN’s instead of their own WWN’s. With VC, HP provides you not only with predefined, HP registered WWN’s, but the customer could use their own from the WWN Private range. So, if a customer decides to change from Emulex to QLogic or vice-versa, the customer is not put into a situation of re-configuring their BOFM settings jus to support that change.
Built-in Management Module Yes No
ITComparison Team Comments IBM offer a hardware management module which fit in a special management slots of the IBM Chassis. It does not use up any Blades slots and does not require any software installation.
HP does not offer a hardware MM, but provide a management software that will require you to install it on a blade or two if redundancy required. It can be installed as well on independent servers. Its disadvantage for HP as it will use up blades slots and require the customers to do installation.

HUH?? Do you even know what a “hardware management module” is? The HP c-Class has the Onboard Administrator which is the brains behind the enclosure management. When comparing the AMM to the OA, it is night and day in terms of functionality and breadth of features. The IBM AMM is not so “Advanced.” Did you even know that only a SINGLE admin can log into the AMM at any time, and if a second admin needed to access a blade or the chassis through the AMM he would kick the other user off of the AMM? How is that efficient management? How is that “advanced?”

The software you are most likely referring to is the HP Insight Control Management suite. This suite of software allows organizations to manage their infrastructure, not just the enclosure. One does *not* have to install ICM just to manage their enclosure(s), and is not a requirement. Heck, it doesn’t even need to be installed on a blade.
Efficient utilization of available power resources
- PowerExecutive™
- Power Regulator, a bit Less functionality and over $400 charge
ITComparison Team Comments HP and IBM power management software are offering almost the same functionality with IBM leading with few enhancements. In addition, IBM is providing their PowerExecutive as freebie where HP is charging for it. As far power consumption go it seems both vendor are doing almost as good and the difference in consumption depend on the configuration ordered by the customer. Most of our testing resulted with power difference less than 5% of the two with IBM consuming a bit less in most scenarios.

With everything else you have posted thus far, I seriously doubt and challenge that you did any testing what-so-ever of the two products. Plus PowerExecutive is just a reporting tool. They have no way of capping today. That is a future “feature.” With the iLO Blade Select or Advanced license (which is included with Insight Control license BTW) you not only have power reporting capabilities, but capping as well.
Investment Protection
- Across Chassis compatibility
- Each chassis is a fully different game
ITComparison Team Comments IBM has been successful in making their chassis totally backward compatible with their older modules and blades and most of their newer modules and blades fit in their older chassis with performance restrictions in rare cases, but that offer a great investment protection to customers who is upgrading their chassis comparing to HP which forcing their customers to toss their old blades and modules out as none of it is compatible across chassis. Who knows if the next HP chassis will follow up the same path as their current one, which mean a total lost of investment when upgrading.

Yes, and this is why IBM continues to lose major market share in the blades arena. They consistently force their customers into limited choices. You want BOFM. Sure, but IBM doesn’t support the LS20, HS20, HS12 or any of their non-x86 product offerings. It’s like if HP stated “hey, we’re going to give you this great technology, but it will not support the BL460 or BL465 first generation product.” Do you think customers would be happy and welcome that technology? I don’t think so. You want to use that older BladeCenter chassis, which provides better power efficiency over the BladeCenter H product? Sure. But, you can’t support the HS21XM, or any other the higher bin CPU’s nor BOFM. There are just too many sacrifices that IBM forces their customers into just so they can continue to spew out “backwards compatibility.”
chrislynch
 
Posts: 1
Joined: Wed Jun 18, 2008 5:18 pm

Re: HP Blades vs IBM Blades Forum

Postby ServerSlinger » Mon Apr 27, 2009 11:50 am

I notice this has not been updated since the Nehalem release and IBMs new support of Hot Swap HDDs on the HS22 blades.
ServerSlinger
 
Posts: 1
Joined: Mon Apr 27, 2009 11:46 am

Re: HP Blades vs IBM Blades Forum

Postby forumadmin » Sat May 09, 2009 9:16 am

Thanks for the feedback ServerSlinger, I will inform our team to update it the soonest.
forumadmin
 
Posts: 36
Joined: Tue Dec 18, 2007 5:52 pm

Re: HP Blades vs IBM Blades Forum

Postby msright1981 » Sat May 09, 2009 9:49 am

I was wondering if Chrislynch work for HP, as that is the same pitch I heard by HP about 2-3 years back. It was such a sad investment & a very bad mistake that our IT manager had lost his job over it. We started with the HP p-Class with over 100 blades. As these started to have problems & the limitation & problem of the p-class chassis started hitting us we had decided to move on to the c-Class. Ah yeah the HP pitch that Chrislynch have been repeated on our IT manager over and over again till he got persuaded that the limitations of the p-Class was not worth it to move the equipment to the new c-Class chassis of course the reason is obvious as they can't do anything. That mean our 200K of blades investment were thrown out of the roof. I believe the IBM approach of being able to move my old blades & modules to the new chassis or use the new modules & blades on the old one with few limitations is a lot better investment than having to replace everything.

Though the waste of this investment was not the reason our IT Manager got fired, but the c-Class chassis still has no real redundancy on power as a failure of one power supply on few of our chassis got 16 blades at a time which was a disaster for a bank operation. I guess some one had to pay for the mess and the IT Manager got replaced. After that our DataCenter has moved to IBM blades. We are running over 200 IBM blades for our operation at the moment including our critical applications & online banking with no problem for the past year. Actually I heard our new IT Manager might get promoted for the stability he provided in the DataCenter.

Other than Power, have you seen any real customers use or adopt Cel? No. Can it offer an edge over the traditional x86 platform for HPC? Only specialized applications, with a long and challenging development cycle could they prevail.


Wow he can not even spell the Cell !!! How many real customers use Cell blades, hmm I can name over 10 oil companies in our region using them in a huge implementation. In addition, they offer a huge performance difference over the traditional x86 platform.

What does the term redundancy mean? More than two. Does IBM offer redundant traces for *EVERY SINGLE PORT*? No they do not. Just because they have two connectors to their backplane does not make it a redundant solution. Ask yourself this. Why did IBM stick with their same backplane design of combining active components (they have over 60 of them on their backplane) and power on the same PCB? If there is even one short the entire backplane is lost; So much for redundant connections.


Wow I believe this was the main problem with our HP blades & seems some one got the points flipped. Ah would it get any better than HP admitting this problem to us after their c-Class chassis started to totally fail after a single power supply fail, yeah they had just admitted to the public. See below what HP said & even on their website.
...This issue is extremely rare; however, if it does occur, the power supply may fail and
this may result in the unplanned shutdown of the enclosure, despite redundancy, and the
enclosure may become inoperable.....
http://h20000.www2.hp.com/bizsupport/Te ... OMER/alert


Well, I don't have a longer time else I would hit more points in there. After testing both in production enviornment IBM blades is way superior to HP blades infrastructure. Ah yeah the new HS22 now over come as well the HotSwap SCSI that HP has always used to persuade customers of their blades. An yes IBM did not implement it over the CPU!! leaving it in a frying pan as our earlier HP blades. HS22 can even support the newest Intel CPU & up to 96GB of RAM which HP has not even offered support for this CPU offering from Intel yet.

I am sorry if I had hit hard on HP, but reading what Chrislynch has wrote I did not want him to fool any one like they fooled us earlier, so please make sure you do you take your choice wisely.
msright1981
 
Posts: 1
Joined: Sat May 09, 2009 9:22 am

Re: HP Blades vs IBM Blades Forum

Postby forumadmin » Fri Jul 24, 2009 7:14 pm

ServerSlinger wrote:I notice this has not been updated since the Nehalem release and IBMs new support of Hot Swap HDDs on the HS22 blades.


Hi ServerSlinger,

Please note we have just updated all the IBM Blades comparison on our site, it took us a bit longer as we were moving them to the new template. Keep coming back for more new comparisons.

Thanks,
ITComparison Team.
forumadmin
 
Posts: 36
Joined: Tue Dec 18, 2007 5:52 pm


Return to Hardware Comparisons

Who is online

Users browsing this forum: No registered users and 1 guest

cron