As stated in this press conference, HP has launched a project called “Odissey” that will probably be a complete game changer in the x86 industry.

So far, only HP-UX could be run on a Superdome, but now, customers will have the capability of running HP-UX as well as Linux in the same Superdome server. The lowest-level virtualization layer of the Superdome is the nPar (node partition) and is an electrically-isolated group of Superdome cells (the picture on the right shows the SD2 enclosure populated with cell blades). As nPars are electrically isolated from each other, it will be possible to have nPars equipped with Xeon CPUs and other nPars with Itanium CPUs. Just as the first generation of Superdomes could run PA-RISC and Itanium processors in different nPars in the same server. A mix of CPUs types or families will not be possible.

Of course, the HP-UX cell blade will need Itanium CPUs and the Linux cell blade will need Xeon CPUs (as Linux is not supported on the latest Itanium-based servers), however, this opens the door to bringing Linux to new levels of availability, making use, for example, of the highly available crossbar of the Superdome 2 that routes all IO signals from the IO extenders, which contain the PCI-e cards, to the cell blades. This crossbar is able to retry all possible transactions and to reroute signals to make sure that every IO is performed accurately.

HP-UX will not be ported under under x86 and it will continue to run on the Integrity blades, rx2800 i2 rack-mount servers, as well as on the Superdome cells with Itanium CPUs. Also, this integration will only be for Intel Xeon processors, not AMD Opterons. The development of HP-UX will continue, as the Itanium roadmap still has two CPUs codenamed “Poulson” and “Kittson” to be delivered in the future.

It would be possible to run Linux (with the current Xeon CPUs – the number of cores of Intel’s next platform, codenamed Sandy bridge, for servers is not clear as of now) on 32 sockets, or 320 cores, or 640 threads !! That is huge and great news for all the customers who wanted to switch smoothly from Unix to Linux, or needed scale-up servers going beyond the 8 sockets provided by most of the vendors.

Also, the Integrity blades, which were very modular (they could be extended from two sockets to four sockets and even to eight sockets by just combining blades together and linking them with a blade link pictured below), will also be made available for Xeon processors.

The new servers (Superdome 2 and scalable blades) are planned for 2013.

Finally, HP announced that the Linux HA portfolio would be similar to the HP-UX one, which means that ServiceGuard for Linux (that was stopped two years ago) will be reactivated.

I think that all these announcements are great news for Linux customers who wanted to push their Linux infrastructures to mission-critical levels. Although HP-UX still has a clear roadmap, the attractiveness of the Xeon processor with Linux on such a scalable and available platform will be very strong.

This offer could also be interesting for customers of other commercial Unix versions by offering amazing scale-up capabilities for Linux on the x86 platform, which is the most open one.

The idea behind CloudSystem Matrix is relatively simple : the whole environment should be as easy to manage as possible.

This starts first with the firmware management. All c-Class enclosures have a defined firmware level according to their Matrix version. This means that the server firmware (HBAs, BIOS, iLO, NICs, etc.), the interconnect modules (HP Virtual Connect Flex-10, Fibre channel or FlexFabric) as well as the Onboard Administrator (the enclosure management processor) have a defined firmware level that was tested and qualified to work together in the best way. Given that HP implementation services take care of the firmware deployment, the administrators don’t have to bother about it.

What can be managed by CloudSystem Matrix ?

The physical servers to be deployed must be HP blades (ProLiant x86_64 or Integrity Itanium servers).The reason for that is that we leverage the capabilities of Virtual Connect to apply network profiles (MAC addresses and WWN) and this technology is available on our blade servers.

However, the virtual machine hosts (VMware, Hyper-V, or HP-UX Integrity Virtual Machines) can be HP blades, HP rack-mount servers (Integrity and ProLiant) and even third-party servers (Dell PowerEdge 2000 series, e300 series , IBM System x servers 6000 series, r800, r900, x300 and x3000 series and IBM blade GS and LS servers) making CloudSystem Matrix probably one of the most open cloud solutions on the market.

In order CloudSystem Matrix to work, the management server needs to discover and manage the targeted equipment. The management console of the VM hosts, the management processors and the interconnect modules must be recognized by the so-called CMS (central management server). It will recognize the presence of the Virtual Connect domain group (which manages Virtual Connect for multiple enclosures) and will put the servers not used as VM hosts as possibly usable for physical deployments.

As soon as the CMS has discovered the equipment, the administrator can use its console on the CMS to create and assign pool of resources to different users.

From this management console, the administrator can manage all the elements provided to both IT architects and business users.

What IT architects need to create their cloud maps first is network connectivity. The VLANs at disposal to the IT architects are the Virtual Connect vNetworks. The administrator provides them to the IT architects using the tab “Networking” on the management console.
There, the CMS communicates with Virtual Connect Enterprise Manager and retrieves all available networks. Each network must then be configured to provide information about the range of IP addresses usable, if the IP address is allocated via DHCP or if the CMS allocates it from its pool of fix addresses.

As soon as a server is put in the enclosure and is managed by Virtual Connect Enterprise Manager, it appears in the “Unassigned” pool of resources. From here, it can be moved to a pool of resources that can be dynamically assigned to a business user. This user will only see the pool of resources that are allowed to him in his self-service portal.

In CloudSystem Matrix, the group of Administrators has all rights, hence they can see all services currently running. The business users can also FlexUp his service by adding either disks or servers to the currently running service, in case, for example, that an unexpected load occurs on the service.

From this console, the administrators can see all items that can be deployed via CloudSystem Matrix: network items, operating systems (retrieved from RDP job, Ignite depots and golden images as well as Hyper-V and VMware templates), storage pool entries, as well as servers. They can control all requests as well as currently deployed services. I will write a new post to explain exactly how the storage provisioning works.

All in all, this third post explained how administrators can, from a single point of control, manage their resources and put them at disposal to the users. The CloudSystem solution is a complete solution that can help IT departments reduce their TCO of up to 56% compared with traditional rack-mount servers. I have already deployed it for customers and must say that many of them are really impressed of the power of the overall solution.

In a long article, Kramer and Porter assert that companies are perceived to be prospering at the expense of the communities. I came to the same conclusion when thinking about an expression of popular culture: TV series. As a child, I used to watch TV series, such as Knight Rider or Airwolf, where private companies or foundations helped fight organized crime and bring justice.
Looking at TV series of today, the contrast is striking. Prison Break or Heroes both stage corporations, which are called “The Company”, probably in order to make them even more impersonal. In their respective series, these companies are instigators of conspiracies and use treason, murder and crime to reach there presumably business objectives.

This expression of the popular culture demonstrates the current perception of companies in communities : ruthless managers will do whatever it takes to optimize short-term financial results at the expense of the rest of the world.

Companies, under customer or regulatory pressure, try to correct this image through periodic social actions. However, according to Kramer and Porter, creating shared value (for both society and corporations) should not be put at the margin of the business model, but at the core.
An interesting example to me was the quarterly earnings calls of a tech company I have invested in. The top managers spoke for one and a half hour about financial results, goals and business  initiatives, until the VP of corporate responsibility had a meager fifteen minutes time slot to present all charitable actions undertaken by the company. Charitable actions are laudable. However, it really gave me the feeling that  making profit while doing something good for communities was an indirect result through the product and services sold but it was not at the core of this company’s business.

I think Kramer and Porter are on the right track and I think Free Software businesses are a great example of creating shared value. Companies, such as Red Hat, Talend, or Pentaho embody the principles of shared value by making billions of US dollars of revenue and supporting communities worldwide. Distributing software under, for instance, the General Public License, these businesses charge neither companies nor consumers for the use of their products, but rather for support, consulting and services. Moreover, they provide the source-code (the instructions that make the programs work) for study or modification purposes.

The authors of the article state that shared value should be created in three ways:

- reconceiving products and markets
The products creating shared value should serve disadvantaged communities, providing products to lower-income homes. The beauty of Free and Open-Source Software (FOSS) is that it can be distributed at no charge. Anyone can install and use a zero-cost operating system and applications, provided by company-sponsored initiatives, such as Fedora or Ubuntu. Of course, communities still need to invest in hardware, but open-source software can be used by poor people in developed countries, as well as by users anywhere in developing countries, thus providing cutting-edge technology (such as virtualization) at no cost.
Moreover, the open-source licenses make it very easy to adapt the products to the needs of users, reusing components already developed by other projects.

- redefining productivity in the value chain
Many of Kramer and Porter’s elements that participate to the value creation, such as “energy use and logistics”, “procurement”, etc. do not apply directly to the production of software products, let alone FOSS.

However, the development of a project, such as the Linux kernel, is a collaborative approach. Anyone can participate and send patches to correct bugs in the program or launch a new module. Companies relying on such community projects and benefiting from the huge manpower provided by these communities cannot take total control of them. Though they can influence them by offering more technological support, they have to take in account the will and motivation of the community to make sure they still benefit from it. They have to balance between their agenda and the motivation of the community. This, to me, is creating shared value.

An example of a bad approach is the OpenSolaris project. After Sun Microsystems had opened and backed OpenSolaris for years as an open-source Unix operating system project, Oracle, which bought Sun, decided to close completely the development process, thus chasing any goodwill to contribute to the project. Although OpenSolaris is still open-source software, Oracle does not allow any change from non-Oracle developers. They do not open their development process and hence do not contribute to any community.

This combination of working for a company and for the community makes me think that employees working for companies selling solutions around Free Software certainly have a sense of working on something greater than just their own business. By helping produce software that can be used to the benefit of anyone around the world, they have a feeling of fulfillment, contributing to the global enhancement of societies.

- building supportive industry clusters
This point is, in my opinion, less relevant to Free Software. The reason for that is that anyone who has a decent Internet connexion can get access to all the FOSS knowledge (by downloading the source code of the programs) and also participate to the process of improving the products by sending improvement suggestions. Obviously, this can be done worldwide – the only prerequisite being a sufficient Internet infrastructure and working computer hardware. The work of NGOs, such as Linux4Afrika may help accelerate the penetration of the market in developing countries by providing support and by teaching classes. Nonetheless, Kramer and Porter name the North Carolina research triangle as an example of a successful cluster. Interestingly, it is where Red Hat’s worldwide headquarters are located.

The HBR article concluded that companies creating shared value would be the ones “unlocking the next wave of innovation and growth”.
I believe companies working in the FOSS ecosystem definitely create shared value. By giving their software for free, these companies make it accessible to all. By using processes based on the Internet, they make it possible for virtually anyone connected to the Internet to participate in their development and support communities. And finally, by publishing the source code, they allow anyone to take a look at how cutting-edge software is written and learn from it.

The current financial success of Free software companies is a proof that their business model is a sustainable success.

“No matter how slick the demo is in rehearsal, when you do it in front of a live audience the probability of a flawless presentation is inversely proportional to the number of people watching, raised to the power of the amount of money involved.”

But here are a few do’s and don’ts that I think make sense to make the most out of your demo time with your customer :

First thing to do : research who is going to attend your demonstration. First, because the technical level of the session may differ completely, depending on who is attending.  An administrator will be interested in bits and bytes stories, a CIO won’t. To make a good use of your (and the audience’s) time, figure out first if they will understand what you will be saying.

Use one card or the other depending on who you are talking to.

Don’t just dive right into the demo if the audience has no background on the technology. When the customer is in front of you, describe the demo environment and briefly explain the basic concepts of the technology, so that everyone in the audience is on the same page. You may want to use PowerPoint slides, a whiteboard, or even a simple speech to carry the message over. End by linking the capabilities of the technology with the advantages this solution brings to the customer. Make sure that these advantages are then clearly highlighted during your demo to make it even more powerful.

One day, I was doing a demo when the customer told me after ten minutes “I am sorry, what are you doing right now?”. I had obviously got it wrong. I hadn’t explained the overall goal of the demo and the customer was completely lost by the actions I was showing.
Tell the customer before you start what the outcome of the demonstration will be. For instance, if demonstrating the capabilities of the HP BladeSystem Matrix, you can say “This demo will show you the automated deployment of a complete IT service – servers, storage, networking, OS and application”. If there are intermediary goals, don’t hesitate to explain them too. Again, here, you can use a PowerPoint slide or a flip chart to list all the important actions you will make and tick them off as you go along, so that the customer understands at which step we are at.

Referring back to point 1), highlight the points which will be interesting for your audience. The administrators may be interested in knowing how things work, hence, don’t hesitate to give the exact description of what you are doing. When I have high-level customers, such as CxOs, I do not explain every technical action I take: they are, in general, not really interested in knowing the technical details. However, if there is anything that brings a business value, such as the capability to charge business users for the IT services they request, I try to emphasize that to them as much as possible.

Regularly ask your customers if everything you are showing is clear. First, it will keep them engaged, and second, it will help you to answer their questions much more easily . It is simpler to
answer a question about e.g. “What happens to the IP address now?” than “What happened to the IP address ten minutes ago when you clicked on the IP button” .  Ask the customer regularly if they are following you.

Finally, try to get an idea of the customer’s enthusiasm level after you are done.  For the solution I presented, I like to ask a question like “how do you see this fitting into your environment?”. Since it is an open ended question, it forces them to think about the current state of their environment and visualize how what you demonstrated would fit in. This can make a big impact on how likely they are to invest in your solution. It’s like when you go to a furniture store to buy a new recliner. The sales guy runs through all the features – the different vibration settings, seat warmer, built in refrigerator etc. but unless you can see it in front of the TV in your living room with you in it, you are less likely to buy it.

All in all, I think that these six steps are key to do an effective demonstration to customers. If it helped you make your customers happy, or if you have your own tips, do let me know in the comments below!

As Command View provides SMI-S connectivity out of the box, it should be easy, right ? Wrong ! (at least in my case).

Usually, you would have a fiber channel host bus adapter connected to the Command View server. However, my CV server does not have one. Also,  the EVA ABM (Array-Based Management), an embedded tool that helps manage EVA arrays, in my case a 4400, does not provide any SMI-S connectivity. The Command View documentation nonetheless states that “If you have layered applications requiring HP SMI-S EVA, you can install the HP SMI-S EVA component on any server that is either connected to the EVA/SAN or has access to HP Command View EVA via Ethernet.”

The hard task was to find how to make this work.

SMI-S provides a utility called discoverer.bat, located in C:\Program Files (x86)\Hewlett-Packard\SMI-S\EVAProvider (yes, Command View only runs on Windows…)

Execute it

Press 1 to add the IP address and the credentials of the ABM.

Verify that the ABM was successfully discovered

Everything ran fine and now, I can discover my array through the Storage Provisioning Manager (SPM), a technology designed to present and deploy storage LUNs automatically as part of BladeSystem Matrix. I will write an entry about it later on !

This Google Summer of Code project written in Python allows to execute a SSH command, to copy files, to rsync directories and to kill processes (among others) on multiple hosts

The install

First download the .tar.gz document
# wget http://parallel-ssh.googlecode.com/files/pssh-2.2.2.tar.gz

untar it
# tar -xzvf pssh-2.2.2.tar.gz
# cd pssh-2.2.2

To install (you need Python, it worked on RHEL 5.3):
# python setup.py install

The configuration

Then to execute parallel commands, create a file with all IP/hostnames you
need (one per line) like this
# cat hosts.txt
192.168.10.1
192.168.10.2

Create the directories “stderr” and “stdout”
# mkdir /tmp/stderr
# mkdir /tmp/stdout

and that’s it !

How to use it ?

If you want to execute the command hostname ?
# pssh -A -e errors -o /tmp/stdout -e /tmp/stderr -h hosts.txt -l root hostname

-A will ask the root password inline
-o will put the standard output in one file per host in the directory /tmp/stdout
-e will put the standard errors in one file per host in the directory /tmp/stderr
-l is the user that will execute the command

For my benchmark I nedded to change the starting script run.sh, so I edited it on one virtual machine and then used pscp to copy it on all the nodes in the directory /root/java/
# pscp -A -h hosts.txt -l root /root/java/run.sh /root/java/

And finally I executed the benchmark on my 32 nodes
# pssh -A -e errors -o results -h hosts.txt -l root /root/java/run.sh

Have you forgotten one JVM garbage collection parameter ? No worries, pnuke will make sure that all java processes are stopped on all servers
# pnuke -A -h hosts.txt -l root java

CloudSystem Matrix is a private cloud solution that aims at speeding the deployment of applications through highly automated technologies and integrated processes.

After we have created and published our cloud map, we have provided our business users all the information they need in order to deploy the application they require to execute their company’s strategy.

In order to do so, the user logs in to the self-service portal.

As you can see, the home page shows which resources are put at disposal to the user, as well as the current service consumption. We see that five physical servers, out of the six at disposal, as well as two virtual machine hosts are in use as part of a service deployed by the CloudSystem Matrix. By clicking on the “Templates” tab, we can access the list of templates which are published and which the user has the right to deploy. Remember the SAP 2-tier template we created in the first part? We have published it, hence we find it back here. That the user has created the template or not, doesn’t play a role as long as he has the right to use it.

By clicking on the template, the business user can display the details of the service: the layout of the network, the price, as well as the notes entered by the creator of the template, etc. In order to deploy the service automatically, he will click on the button ‘Create service’.

The user now has to enter a service name (for instance “SAP 2-tier John Smith”), the hostname completion (remember the need for a hash sign in the first post ?), an email -to contact the service requestor-, a start date and hour at which the service will be deployed, an end-date, at which the service will be rolled back, and finally, a pool of resources, from which the resources, such as servers, storage and networking, will be picked from. Final step: he clicks on “Submit”, and here we go !

The request is sent and the activity which is happening behind the scenes is displayed: the resources are reserved from the pool of resources. The software makes sure that the pool can accommodate the need indicated in the template. We can see here that four servers, four IP addresses, three networks and four boot disks are requested. CloudSystem Matrix will pick from the pool of resources the servers that meet at least the requirements of the template and will pick from storage disks (LUNs) that are already presented to WWN (worldwide names) and also meet the size indicated.

We see here that CloudSystem Matrix also integrates nicely into the user’s processes: the request above is paused, because an administrator (or an IT manager or anyone else who should take this decision) must give his approval, so that the deployment can continue. A Request for Change (RFC) could also be triggered at this point in time with tools such as HP Service Manager or BMC’s remedy. The teams using these tools will just have to accept the request and the deployment will go on.

We see here that the administrator granted the authorization to deploy this service and that the allocation process has started. The requested resources are then provisionned. The physical servers are reserved and a custom HP Virtual Connect profile is applied to them. This means that we are applying one ore more MAC addresses for the network, which will be connected to the VLANs we need, as well as one or more WWN for the storage, to which disks are already presented. The zoning of the storage was already prepared, so that after the WWN is attributed to the server, it can install the operating system as well as the application straight after the VC profile was attributed.

Taking a look at the Onboard Administrator, the HP BladeSystem enclosure management processor, the server loses the “i“-icon, which means that the server has now a Virtual Connect profile

On the screenshot above, we can see “Provisioning the logical server” (either physical or virtual) and, finally, “Activating logical server”: the ProLiant or Integrity server is now starting.

If VMware or Hyper-V virtual machines were to be deployed, CloudSystem Matrix would pick the right template, clone it, provision it, and finally adapt it according to the needs of the cloud map. If HP Integrity Virtual Machine were deployed, the host would be contacted vis SSH, it would create the VM and deploy it with Ignite UX either through normal package deployment, or via a golden image, a concept similar to the VMware/Hyper-V template.

We see above that the Integrity server starts, it then boots from the network (since the disks we have presented are empty) and that Ignite UX detects a deployment request from the client.

All the deployments methods (Ignite-UX,  HP RDP, HP Server Automation, etc.) and CloudSystem Matrix can communicate with each other: the deployment server confirms to the Matrix operating environment that the deployment request of a given client was received. This is acknowledged by the management server which indicates it to the user.

Finally, after the operating system is deployed, the application is installed through the execution of a workflow (it will be the subject of another blog post in the future).

Although I discussed the technical details of what happens behind the scenes, the business users, when they make a request, can see a completely abstract high-level view of what is going on. They don’t need to know the details, they just want to know if the application was deployed.

On the picture above, the business user sees that everything went well and that his IT service is up and running. What could have taken so much time to deploy, lasted roughly one hour with HP CloudSystem Matrix.

Instead of taking months putting all the pieces together, CloudSystem Matrix orchestrates the provisioning of storage, server and networking, installs the operating system and deploys the application in an automated way, also integrating in the organizations’ processes.

It is a fantastic tool that was already embraced by customers to modernize their IT environments. I have personally helped some of them deploying this solution in their datacenters and they love it. Though it needs a good understanding of key HP technologies (HP BladeSystem, HP Virtual Connect, HP Systems Insight Manager, etc.), the HP services team take care of the deployment of the solution, and the complexity of the infrastructure is hidden to the business users.

In my third and last blog entry for this series, I will focus around managing the infrastructure to fill the different pool of resources and how to present them to the users.

During the last past years, the rise of Linux-based operating systems, mostly on x86 architectures, was spectacular, because distributions such as Red Hat were able to provide a lower-cost alternative in terms of acquisition. However, the vast majority of these decision-makers, who, besides, also embraced Linux in their datacenters, thinks that Unix-based operating systems still bring more value in the following categories:

- availibility

Indeed, when it comes to availability, Unices have an advantage. The maturity of the product is higher, but above all, the integration of the platform to the operating system and to the application give an indisputable advantage to RISC/EPIC-based servers. The clustering solution ServiceGuard, for instance is way ahead of what SuSE HA or the Red Hat Cluster Suite can dream to become in 2 years. Though it is possible to reach high levels of availability for the database stack on Linux with Oracle RAC (Real Application Clusters), the management of the solution is quite tedious and its cost prohibitive.

- scalability

Clearly, the biggest deployments of very large databases or ERPs happen on Unix. However, Linux is catching up very fast. Red Hat 6 supports the 120 logical CPUs and 4TB of RAM of the brand new HP DL980 ! Especially when it comes to scalability with regards to virtualization for massive consolidation scenarios, Linux does an excellent job with the compelling argument of the full binary compatibility with the x86_64 architecture.

- ease of maintenance

The number of updates is far less frequent for Unix (at least for HP-UX that I know well) compared with Linux. Moreover, tools such as Dynamic Root Disk, which allows to relocate the operating system from the local disks to the SAN in one command line and one reboot, or to patch the OS on a mirror offline and to reduce the planned downtime to the reboot time, really increase the productivity of the admins. Once again, the tight integration of the operating system to the platform makes this happen.

- partitionning

The partitioning of the RISC-EPIC hardware platform is much more advanced. Because of the capabilities of their platform, x86-based servers rely mainly on hypervisor-based virtualization, whereas Unices offer both hard and soft partitions. Also, although IBM offers Linux on Power servers, most of the virtualization capabilities are only available for AIX, IBM’s Unix. Finally, Linux lacks the pay-per-use and workload management functions of Unices. cgroups implemented in RHEL6 are a beginning, but it still lags behind the HP SRP or Solaris Zones.

A personal note about the “Staffing and skills are not an issue”: my first system administration experience started with Linux and I learned HP-UX afterward. The main principles of *NIX, such as LVM, runlevels, init scripts, the Bash shell, etc. are the same and hence the lack of smart people able to learn a few more commands should never appear. The most important is to stay flexible and ready to learn new technologies !

To conclude, Unix is going to remain in the datacenters for long. Its stability, ease of maintenance and workload management ensure a great continuity in the IT service delivery. Though the number of Linux servers will continue to grow, driven by the increase in performance of the x86_64 platforms (and I am definitely fine with that !), HP-UX, AIX and Solaris (depending on Oracle’s good will) are here to stay…