Being from the services side of HP, I don’t always know some of the other areas of HP as much as I should so I had a discussion with Doug Oathout about Converged Infrastructure (CI), and asked him some questions about CI in hopes that I’d know more when I talk with clients. The following is a summary of what Doug and I discussed – filtered through my ears…
- Is Converged Infrastructure really the next big thing and if so, why?
Yes, it is the next big thing. The reason why is because IT (over the last 20 years) has been put together as a set of projects or silos and once deployed remained relatively static. That’s caused most organization to be overprovisioned (wasted computing power) . The reason CI is the next big thing is because the applications can share the infrastructure more effectively. The applications can live and breathe. Smaller increments can be added or removed as needed.
The converged part of CI is bringing the server, storage and networking hardware together so that it can be broken up into smaller increments. The converged infrastructure also has the management software so that the applications can be moved around as needed to meet the service levels of the business. Applications are now matched to the infrastructure and that is also matched to the needs of the business.
It is enabled by the software and the management methods that are used, and the hardware infrastructure has been tweaked as well. All the other players in the industry are a network player or a storage player and optimize their stack. The infrastructure between them are ad-hoc, with little optimization. With HP’s converged infrastructure it’s all been optimized.
For example: HP’s technology Virtual Connect takes the virtual I/O capability of the server and creates a flexible fabric that can connect to any part of the infrastructure (storage, networking…) though one connection/pipe. That pipe can be divided up as needed as well. Being able to optimize between the stack allows it to be simple and wired once.
The same thing can be said about the optimized environment. It is one resource pool. You can take processing on line for a process as needed and remove it as well. This enables an incredible increase in productivity and simplification, removing unneeded connective tissue.
- This is starting to sound like an infrastructure as a service, private cloud enabler. Is that right?
Yes, it definitely is. Allowing for running Exchange or SAP in an optimized fashion.
For example: We can take a 3-tier SAP implementation and use one of our blade systems, with some parts on HP-UX, other parts on Linux and even windows. The resources can be moved between these environments as needed. One connection hooks it all up. Allowing a much more usable and efficient environment.
- Why is CI important and to whom?
As clients are modernizing their infrastructure they are looking to optimize their costs. A converged infrastructure is cheaper to run long term. It allows for more flexibility as well (moving resources in and out of an application’s environment). It more efficient to buy and operate.
- What CI issues are customers worried about (Vendor lock-in, etc.)?
There are a couple of issues that customers need to think about when moving to CI:
- You’ll need to change your processes. With the matrix operating environment or with HP’s orchestration operations, we can bring applications on in minutes. There is a need for a process for who can add (or remove) processes. Today most organizations have a long workflow to make these changes with multiple signoffs. When resources are no longer needed, the environment can take resources off line as well.
- People have siloed jobs today where the network, server and service manager all live in isolation and control their environment. That will change -- with resources coming on and off-line they need to be aware of the implications and procedures of others. They need to talk with each other. Now that the server manager may be dialing up and down the virtual connect ports on the server they are not impacting the network and keep the information up-to-date on the configuration. Similarly now that we have iSCSI and fiber channel running across Ethernet, we need to let the storage manager understand what storage needs to be connected to what virtual I/O port and need the needs of the application. The jobs will need to evolve a bit to include a more holistic view. The expectations will need to change.
- In the area of lock-in, some people do have concerns that they are going to be forced to use HP software and hardware from now on. All of HP’s CI interfaces are based on standards. All of it is built on open integration. It works great in an integrated environment but still works with Brocade, Emulex, EMC, Cisco… and has been delivered that way with many customers. It is not a lock in strategy, we just view that you’ll lose some of the benefits when you start to mix in other parties.
- What are customers saying about CI? The pros and cons?
It’s clear that clients need to define an architecture that allows them to automate their processes. Clients can usually identify the 30-40% of applications that can easily move to these more flexible environments. The next 30% or so are a bit tougher, since they have unique requirements. The last 30% are really hard to move.
HP has a number of offerings addressing this final hard to manage area:
- Outsourcing… Let someone else worry about it. It doesn’t matter if it is AS/400s or the mainframe HP’s Enterprise Services will manage it, on your site or at an HP data center.
- Application modernization is another area where architecture analysis is needed. I’ve blogged quite a bit about apps portfolio and modernization and the various levels of porting, reengineering or development based on the business requirements. It can be brought back in-house at that point if desired.
- How do you see CI generating new value within the client’s business?
When you can get to a converged, fully enabled CI, you can more quickly bring web servers, business process… up in a short time. Its direct effect on time to market generates value.
If you want to start up a new business model: moving to more web sales, the change can be made quickly. It also allows you to grow and scale as needed.
- Does HP have a mechanism to allow clients to have more infrastructure capabilities on-the-fly?
Clouds on-line with our matrix will allow for a number of CPUs for a short duration. Development and testing environments are a perfect example of where clients are scaling up some hardware “by the hour” to address a business need. DreamWorks is another good example. They connect to HP for about 20% of their rendering needs for short term rendering. Once you get the core infrastructure converged, you have additional flexibility so if you want to change the response time or power consumption for certain transactions for a while you can dial it back or scale up as needed.
- What are the 3 things you want people to know about converged infrastructure?
- You can place any applications, anywhere across your data centers once CI has been implemented. It could be in Atlanta, California or China - quickly.
- It optimizes all the time to meet the business requirements on-the-fly.
- It is really about wiring the infrastructure up once and then configuring it. If you want to change the performance of an application, you don’t need to rewire the environment – it’s all done in software.
I just noticed there is a CI blog as well. I guess I’ll have to start reading it too.
