HPE BladeSystem Reviews

I really value the FlexFabric interconnects.

HPE BladeSystem was introduced by me as an architect to boost the performance/server footprint, especially with VMware virtualization.

The storage blades could be improved.

I have used it for eleven years.

There were stability issues in the early versions, Blades G1.

I did not encounter any scalability issues.

I would give technical support a rating of 8/10.

We used rack mounted servers.

The initial setup was straightforward using a wizard.

The customer has to decide and evaluate the tradeoff between CAPEX and OPEX.

We evaluated the IBM Blades System.

Examine your infrastructure KPIs. This will typically include analyzing a reduction in OPEX, ease of operation, ease of troubleshoot, decreasing cabling, and increasing footprint performance.

HPE BladeSystem c7000 is a complex piece of engineering.

• I appreciate its simplified interface that gives the administrator the power to build a complete processing infrastructure remotely.
• I like the on-board administrator as it gave me a very detailed set of information that allowed me to manage every aspect of the infrastructure remotely.
• It allowed me to centralize the iLO remote access to every blade in the cabinet with very good performance.

You can view a Bladesystem like a modern car. The first thing you see is the body and the glossy paint, but under the hood, a bladesystem is essentially a group of servers (multi-core processors, RAM, Buses, storage, etc ), with redundant variable speed cooling blowers, redundant power suppliers, a large set of redundant connectivity options, and a big quantity of temperature and power consumption sensors, all of those connected and administered from a redundant administration module with many configuration parameters that you can accommodate in a bunch of ways to satisfy many different requirements.

Everyone of this modules are an appliance (a complete computer itself), and you can have a duplicate of the OA and the VC, just for redundancy and high availability purposes.

Before this brief description, probably you would agree that this is a complex architecture.

But the most attractive part of that , is that you deal with this complexity through a web portal that concentrate all of the configurations options, easy this tasks and guide the user with several wizards.

To manage all the parameters related to the Enclosure or Chassis, security access, and monitoring, you have to enter to the "onboard administrator module" (OA).

To manage all the aspect about LAN or SAN connectivity to the server blades you have to jump into the "virtual connect Module" (VC), but don't desperate, you have an hyperlink from the OA, that open the VC portal, to give you a seamless navigation between modules.

At last, but not least, you have the blades servers itself. You can have up to sixteen of those servers, with processors, memory, Out of Band management processor (Insight Lights-Out or iLO) and I/O cards (NICs, HBAs, CNAs, etc)

All of those have several Firmware (BIOS, NIC firmware, Power Regulator Firmware, HBA Firmware, iLO Firmware, Onboard Administrator Firmware, Virtual Connect Firmware, etc ) and you need to solve incompatibility issues between all of those.

The best part is that HPE give you an utility (HP Smart Update manager) that can manage all those firmware in a consolidated way.

HPE works hard to provide a centralized administration and good experience with the software, and if you are an advanced user, also can use an add-on to access all the configuration parameters using powershell (the administrator's task programming language that come in every windows operating system).

The first goal was to use blades. This stemmed from a space problem in our data center. We needed to add more servers, but the space became short quickly. The first consolidation approach was a blades server.

• We administer all the systems remotely.
• The blade server standardized more of our configuration processes with less manual intervention.
• We also found that we needed less cabling. Now we can connect 16 servers in 10 rack units with 3 LAN and 2 SAN Fiber, instead of 48 LAN and 32 SAN fiber.

• Hardware management could be improved.
• Cisco UCS has a more universal approach. It treats the hardware as stateless and manages absolutely all configurations from the same console.
• HPE has an on-board administrator to manage hardware aspects and virtual connect to manage LAN and SAN connectivity between blades and the rest of the world.
• Firmware updates are complex. There are so many components and you need to account for the compatibility of all parts. Otherwise, you can have a blade that cannot start with the new firmware and then it takes extra time to solve the problem.
• HPE has a utility for firmware updates that tries to provide peace of mind. It takes all these variables into account.

We have been using this solution since 2008. We began mounting an 8 node VMware cluster. We began with one enclosure, a cabinet with 16 blade servers. We now have more than 18 of them distributed in different locations around the world.

We did not have any stability issues. The quality of the server itself enhances stability. Once the server is running, it runs for a long time.

We don’t have issues with computer power scalability. We just add more blades, configure, install and go, or add more memory to an existing blade.

HPE also supports mixing several blade models in the same cabinet.

You can have, for instance, BL460c G7, BL460c Gen8, and BL460c Gen9 working smoothly in the same C-Class enclosure.

In my country, the level of support is quite good. I recommend that you buy the server with a three or give year care pack to receive the manufacturer’s warranty.

I used rack form factor servers and switched to blades to gain consolidation ration. I also wanted to have better management control over the hardware infrastructure.

The setup is easy if you install only one cabinet, and you know what to do and what to expect from the platform.

When you plan to grow your infrastructure to more than 16 blades, it becomes a little bit complex. You need to think about how to manage Virtual Connect Domains, MAC virtualization, and WWN virtualization.

If you design your platform based on that, everything will go fine. You will know what to do when a problem arises.

We have only used HPE infrastructure. Previous to blades, we were using Standalone Rack form factor servers like the DL380 model.

If I were brand agnostic, I would probably select CISCO UCS, but this didn't exist when we decided to use HPE blades.

Now, with Synergy Composable Systems, HPE probably takes a leap forward in technology and puts itself at the forefront. Please keep in mind that technology.

• The product is good and strong. Nowadays, the software works fine. I chose HPE because of the existing vendor relationship and reputation.
• You can do all of the installation and configuration tasks. However, if you are not experienced, contract your first installation service to a partner with a lot of experience in that kind of equipment.
• Take the corresponding training, as it is very useful.
• I recommend taking the official HPE platform support. They have a BladeSystem course.
• I also recommend users strengthen their knowledge with a Virtual Connect course. With this in mind, you can have a good experience with this kind of platform.
• I also recommend that you progress further and think about automating the procedures, installations, and decommissions as much as possible.

• Allows immediate access for server management
• Allows immediate detection of the access logs
• Secures access to the console
• Provides reliable administration

• Enabled remote management of the equipment's problems/issues
• Helped in remotely identifying the power to review the status of errors

I would prefer to have changes in the compatibility of the blade servers along with the new ones designed by HPE. The superior equipment does not allow this at the present time.

I have used this solution for a total of eight years as the infrastructure manager. Currently, we are using the HPE BladeSystem c7000 enclosure and the BladeSystem c7000 Onboard Administrator Tray 1.7 enclosure devices.

In the past, we have used:

• Server Blade ProLiant BL680c G5 HPE 2.12, Jul 16, 2012
• Server Blade ProLiant BL680c G5 HPE 2.06, May 31, 2011
• Server Blade ProLiant BL460c Gen8 HPE 1.20, Feb 01, 2013
• Server Blade ProLiant BL460c Gen8 HPE 1.20, Feb 01, 2013
• Server Blade ProLiant BL460c Gen8 HPE 1.30, Jul 18, 2013
• Server Blade ProLiant BL460c G7 HPE 1.20, Mar 14, 2011
• Server Blade ProLiant BL460c G1 HPE 2.23, Nov 05, 2013
• Server Blade ProLiant BL460c Gen8 HPE 1.20, Feb 01, 2013
• Server Blade ProLiant BL460c Gen8 HPE 1.20, Feb 01, 2013
• Server Blade ProLiant BL460c Gen8 HPE 1.30, Jul 18, 2013
• Server Blade ProLiant BL460c G7 HPE 1.65, Nov 08, 2013
• Server Blade ProLiant BL460c G1 HPE 2.06, May 31, 2011

We have not experienced any problems with the implementation of the product.

When cases were opened, we received very good responses to the issues. Escalations were also done in a timely manner.

I have not used any other solution.

The solution was implemented by the provider. It was done in a simple way.

It has not been applied from a remote site, only from the current infrastructure. You should have a personalized follow-up for the management/administration of the implemented solution.

In regards to the prices, it should adapt to the current needs of the country. Granting of the licenses has always been done in a prompt manner.

Here are the valuable hardware and software features:

Hardware:

• Multi-generational blade system which supports all the way from Gen1 to Gen9 without a forklift upgrade or keeping old enclosures around.
• Up to 16 servers of varying physical sizes (half and full height) allowed in a single enclosure.
• Multiple scalable backplane options (including 16Gb Brocade fiber switches and 10Gbe Cisco Fabric Extenders) to ensure optimal redundancy and interoperability with other datacenter hardware.
• Allows the use of network defined architectures without the need for additional training.
• Allows storage admins, server admins, and network admins access to each component without additional cost or delay due to lack of familiarity.
• Multiple power options: Single and three phase power for different datacenter layouts.
• Versatile with optional virtual connect module to enhance east/west traffic flow for virtualization, reducing unnecessary north and south traffic.

Software:

• Onboard Administrator
• Insight Control
• HPE OneView

This solution has improved our organization as follows:

• Ease of management and configuration through a single pane of glass for up to 16 blade servers.
• Leveraging the Onboard Administrator for control over troubled servers without the need for physical access to the servers.
• Ability to reset blade server fuses remotely if they become completely unresponsive. Removes the need for many calls to a datacenter for support.
• OneView has taken this a step further with newer generation blades. This cannot be leveraged on the blades older than Gen7.

The Onboard Administrator (OA) is the heart of C7000 enclosure management, it integrates advanced health and monitoring to the point of even iLo launching through the OA to KVM a blade server that is not responsive. There are two OA modules, one active, the other passive.

HPE contacted us after identifying the serial number of our active OA module was in a critical recall list.

While all components are redundant including two Cisco Fabric Extenders (Fex) for Network Connectivity of the blade servers (Our config is 4 x 10 Gbe per Fabric Extender teamed with the second Fabric Extender through Virtual Port Channel). A single OA replacement caused 15 servers with VPC protection to be dropped off the network when the device was failed from active to passive; completely disrupting 200+ VMs and >2000 users; all when it was assured to failover gracefully.

If I was to physically disconnect all cables in one Fex, or even remove it from a blade enclosure it would operate fine, but for some reason, the bad OA module caused a network reset to occur which eliminated all redundancy on the link.

For more information on VPC to elaborate it’s power, flexibility and resilience, please see the following links: http://www.cisco.com/c/en/us/p...
http://www.netcraftsmen.com/ho...

We have been using this solution for nine years, through different revisions.

We only had a stability issue with a defective OA Module. Other than that, most blade servers have only seen odd failures on hard drives or RAID batteries on older generations.

The super capacitor in newer generations seems even more resilient. No critical outages have occurred at a blade enclosure level in over nine years of use with c7000 generations, other than a single defective OA module.

There was a legendary update on the chassis itself, in a mission critical ERP/Citrix environment.

We have never reached a threshold on scalability with the c7000. With multiple I/O interconnects on the backplane, we have never maxed out its capabilities.

I recommend having Proactive Care Advanced Support or Datacenter Care on any critical datacenter infrastructure. When leveraging this level of support, we have had great results hitting the advanced resolution center.

The Proactive Care and above coverage also grants you a TAM and quarterly health checks including intimate knowledge of your environment, interoperability of your firmware versions, and recommendations to avoid potential conflicts.

The tool was up in two hours, fully migrating blades from old to new c7000 included to this timing. Fiber zoning and ethernet configuration is a breeze. As long as you know your products from the last generation, it’s quite intuitive.

Always ensure that you purchase the Insight Control Suite. This will ensure all current and future blades are licensed for iLO and other features. This will ensure ease of use and configuration.

We evaluated IBM BladeCenter and HPE BladeCenter M1000 Series.

It is quite easy to setup, configure, and get fully into production. This is the case even when physically migrating old blades from an old c7000 to a newer c7000 chassis. That can happen in less than two hours.

When placed on the same fiber fabric and properly utilizing aliases, these are completely hot swappable between chassis. This is super-efficient and smooth on the transitions and decommissioning of the older chassis.

Be warned: Even though G1 servers are supported, having a 2Gb fiber card in the old G1 servers will cause them to be non-functional with new 16Gb fiber switches. This is not an HP issue, but rather a limitation of legacy fiber. Plan appropriately when upgrading older technology!

• Easy to add new blades
• No additional cabling required for network or FC
• Nice single point of management
• Easy portal

It has given us a much reduced server footprint with a subsequent reduction in cabling and a lower carbon footprint.

• Ability to zoom in on the blade system graphic
• Easy way of reporting all the blades at once

This has been in use for three years.

In three years, we have not had any issues with enclosures nor with server instability.

We have had no issues with scalability. We have added ProLiant servers to the chassis which shares storage.

Technical support is great, once you get to it

We previously had rack-mounted HP-UX servers. Our hardware refresh dictated that we move to blade systems in order to keep using HP-UX systems.

It was a very straightforward migration.

Ensure that you use virtual connect modules. It is well worth the extra cost over traditional switches.

Always go for virtual connect modules from the outset.

The virtual connect side of networking and the manageability through that is by far the biggest win for us. The blades come and go as racks do, but the virtualization back of it means a lot less hands on and a lot more manageability.

The biggest benefit is the minimum downtime due to the programmatic nature of the whole thing.

There’s nothing that I don't already know is coming out.

We've had them for quite a few years now. Early on it was a bit hit and miss, but more recently it has become far more stable.

We have not gone full scale with it. We only have it in small areas of the data center at the moment.

We have not used technical support.

We knew we needed a new solution because our data center costs were rising on our racks and we just had to slimline down into a more compact solution.

The early stages weren't as smooth as they should have been. I was involved in the initial setup and it was complex because of the nature we wanted to use it in; a very virtualized network and storage capacity. It wasn't quite straightforward and it meant a great deal of complex planning to make sure we got it right in the first place and the initial setup didn't cause problems later on.

We also looked at Dell solutions.

When selecting a vendor, reputation and pricing are most important.

Spend as much time as possible planning before you go anywhere near it.

I get the same level of service and perspective for everything. It is a generic experience with respect to what it's used for.

I would like to see all-around fulfillment availability from top to bottom in terms of features. I guess the answer is higher availability.

The stability is pretty good. I'd say it's on par with other vendors in the same category.

It scales fine for tens of thousands of things.

I can’t comment on things that I'm not supposed to.

We were using a multitude of vendors with comparable differences.

We considered the typical ones, the big three: IBM, HPE, and Dell.

I needed a solution because of my economies of scale. I had the demand and a business need.

I would advise colleagues not to rush into settling on a solution. When we were choosing a vendor, we considered price, performance, availability, and engagement.

The BladeSystems come with 16 blades in one chassis. This gives us a central location to manage all our HPE blade servers. We already apply a virtualization hypervisor layer on it, like Microsoft.

I would especially like to see a hyperconverged solution from HPE. With technology becoming faster and faster, and everything going to the cloud, I think cyber convergence is very good.

I’m fine, you know. I think there is no need to improve it.

Honestly, it's very stable. There have been no issues with it during the past five years. It's very, very good. That’s why we would like to expand our data center now to buy many enclosures to accommodate the demands of what we have right now.

It is scalable. I think that it is like HPE Virtual Connect Enterprise Manager Software. You can configure the network from a central location for all enclosures. We have more than four or five enclosures. That is very good for us. 

That's why I would recommend having one stack of vendor products. For example, we have HPE blade servers, some network devices, and some switches. This way, you can integrate and monitor the health status for all enclosures from a single monitor. You are alerted if any failure occurs. That makes our life easier in IT; so it's good.

I wasn’t involved in the initial setup, but my colleague says that it was good.

We compared HPE blade servers with Cisco blade servers. Cisco is good, but expensive. I don’t have experience with them, but one of my friends has it. Later, they decided to go with HPE. 

From our perspective in IT, you should look for the solution which fits your needs. It doesn't matter who the vendor is; but if you go with HPE, they have a lot of good experience in data centers. 

We decided to go with HPE blades and enclosures. They are very powerful and they fit our needs. HPE fits our needs, especially because they also have the back-end storage, like 3PAR. 

I recommend having a full stack with a single vendor, so you only have to deal with one company for support.

Now our company is thinking about buying new servers. It's called HPE Apollo, I think. It's in progress. It's very good. I hope we will get more experience with HPE, especially from a training perspective.

We were looking for stability. It's very powerful. It has huge resources. You can customize it as you like, and it fits our needs. Of course, it takes up little space. It's one rack in which you can put three enclosures. You can build your entire data center in one rack. It's very good. 

Since I joined my current organization four years ago, HPE did an amazing transformation of our data center. Everything there is an HPE product from A to Z. It's an end-to-end solution, including the hardware, networking, software, and everything. We are very comfortable with it. In five years, we haven’t faced any issues with it. It's very good.

We migrated our entire legacy landscape from old HP Superdome infrastructure to the new x86 infrastructure on blades. In terms of application portability, it was a seamless project from my side.

The main benefit has been in terms of the newer platform having much more computer power. You can see your workloads completing much quicker. Users with feedback in terms of workload performance have reported that a job that typically took 3 or 4 hours to run took a few minutes.

We're busy with phase 2 of the project. We classify the legacy landscape into simple, medium and complex applications. We've done the simple and medium. We're now going to tackle the complex and will make use of HPE professional services to see how we can either modernize the application code, for which we don't have the skill set anymore, or some subset of the applications.

In terms of platform stability, the legacy infrastructure was out of warranty and out of support. The new platform was much more stable and reliable.

We have not used technical support often. HPE worked closely with us on the project. It was a joint initiative. We never had any issues and customers didn't even know they were migrated.

We're a telecom company. We've standardized around HPE, IBM, Oracle Solaris Sun boxes and now we've got VCE Vblocks. In terms of percentage, HPE's still about 60% of our landscape.

We moved to an x86 infrastructure and we could probably move other workloads from IBM-related infrastructure across to an HPE infrastructure. We could standardize further, but from an x86 perspective, it becomes more or less agnostic regarding which vendor infrastructure is the underlying infrastructure; as long as the operating system and virtualization can take place on top.

It did take us quite some time to make the business case and to get the funding, but once we got all of that, the project went smoothly. We never had any issues.

For the technical part of the setup, we brought in professional services from HPE and they did the initial configuration and the racking and stacking of the systems.

We were looking at VCE Vblock, HPE, Oracle and IBM. We went with HPE because the HPE platform is about 60%, IBM is about 10%, and Oracle is about 30%.

We're based in South Africa and we're pretty much dependent on the OEM to help us and advise us. HPE is able to advise us, especially on the architectural designs they can actually vet the architectural designs and give input that has proven to be valuable to us.

We always try to check supportability of the platform post-project implementation. Trust me, before the project implementation, before they kick up, all the vendors will lobby. Obviously, when you’ve gone live and you start having problems after the fact, the vendors stay far away. So you need to make sure that in-country support is there and the skills are there. For example, we had an experience with Oracle where they didn't even have the skill sets at the right support levels.

For us, the best feature of the BladeSystems is the plug-and-play capability. We are able to replace one quickly; everything's hot-pluggable. The enclosures are well made. The way we manage our enclosures, everything's done through one iLO. It's really easy to keep on top of things. They are what they are. They do the job they're designed to do.

When we initially went with blade servers, our footprint in the datacentre went down. We still have to use large ProLiant servers. We mainly use it in a Citrix environment, so we use each blade as a Xen host. Although we might have about 400 blades, we've then got four VMs running off that, whereas in the past, we would have had thousands of servers all stacked up on top of each other.

We have a niggle with the iLO connection. Sometimes you can go from one PC to another and how the iLO reacts depends on the web browser setup. I'd like to see something like HTML5 setup for iLO, so you don't need to install any Java or any of the plug-ins; so it's completely universal.

It’s a great system, but it’s not perfect yet.

We've had blade servers for about 10 years.

We don't have any real issues with stability.

We have a HPE support contract. If anything does go wrong, it's just, swap it out, put it back in. I'll be careful what I mention, but sometimes the technical support online isn't great. You're often dealing with people who just ask you basic questions and ask you to check things that you've already checked. You know the fault's there; you wouldn't be logging the call if the fault wasn't there. But the engineers that come on-site are great.

Initially, we were using Compaq equipment. Then Compaq and HP merged. We did do trials with IBM, but we found it to be a bit of a step back in that the kit just wasn't quite as high tech. It just seemed at the time it was a massive step back in the way we were working. HP seemed like it was a step ahead.

My advice is to invest early dollars on higher capacity. In the early stages, we didn't put enough high power kit in and we ended up upgrading all the time, especially on the memory side. I think we went in quite low with RAM and we went through a massive process of upgrading all that.