About Me

My photo
Scott Arnett is an Information Technology & Security Professional Executive with over 30 years experience in IT. Scott has worked in various industries such as health care, insurance, manufacturing, broadcast, printing, and consulting and in enterprises ranging in size from $50M to $20B in revenue. Scott’s experience encompasses the following areas of specialization: Leadership, Strategy, Architecture, Business Partnership & Acumen, Process Management, Infrastructure and Security. With his broad understanding of technology and his ability to communicate successfully with both Executives and Technical Specialists, Scott has been consistently recognized as someone who not only can "Connect the Dots", but who can also create a workable solution. Scott is equally comfortable playing technical, project management/leadership and organizational leadership roles through experience gained throughout his career. Scott has previously acted in the role of CIO, CTO, and VP of IT, successfully built 9 data centers across the country, and is expert in understanding ITIL, PCI Compliance, SOX, HIPAA, FERPA, FRCP and COBIT.

Monday, December 12, 2011

Build, Rent, or Cloud Services?

I had a colleague call me a few weeks ago and was seeking advice on a data center strategy. Their data center is 25 years old, the environmental controls need replacement, they need space – do I think they should remodel and expand. We talked for a few hours to get more information on the current state, desired state and future state. During the conversation it became very clear that there is confusion between co-location options, Cloud Services (SaaS, IaaS, etc) and internal options.

My colleague said that there is confusion out there and I should put this out on my blog as others may be asking the same questions. I was more than happy to oblige, with one condition – that they read some of my cloud posts. So we have a deal, and here we go.

My first recommendation to my colleague was to perform an assessment of what they have today. This assessment should include the facilities, but also, networks, servers, storage, tools, applications, access options, capacity, and disaster recovery. Once we have that completed assessment, a picture of what we have in place today, let’s identify today’s pain points. This quickly revealed that it truly is a facility issue that is putting constraints on the operations, and the ability to deliver capability to the business.

To remodel a production data center online is almost impossible. I have done it once in my career, but it comes with high risk, and many challenges. In addition, to make that investment of building a new data center and make the TCO financially sustainable, you have a great deal of homework ahead of you.

So to make sure we are all on the same page, my definition of a co-location is taking your operations and renting space from Joe’s Data Center and putting it on their floor. You pay to rent the floor space that includes power, cooling, and network / internet connectivity. It is your servers, storage, equipment. You still maintain your process(s), procedure, operations, monitoring and break/fix. A hosted solution is just moving your application(s) and data to their data center on their servers/storage and you maintain the application, they maintain the infrastructure. Cloud based services is renting an application or called Software as a Service (SaaS) or renting some storage for DR or called Infrastructure as a Service.

There are benefits to each scenario and you have to look at the cost(s), risk(s) and operations. In addition, your disaster recovery plan. Going to an option that takes your mission critical infrastructure and applications off premise comes with risk. You have to take into account carrier performance, geographic risks, power grids, and so forth. If your corporate office or key production facility just lost internet connection, they no longer have access to applications or infrastructure – what impact does that have on the business? Do you have redundant circuits between different carriers? Are the different carriers all renting space on the same fiber that was just cut? Do you have redundant power grid supply lines – from different substations? You now have all these factors to consider as your data center is miles away from all your users and many things out of your control. Here is a diagram I found in some of my archives:

 This diagram shows connectivity to the primary data center from multiple facilities with point to point connections. There is new technology out there to utilize and investigate, such as MPLS. You can also push down to the client to determine which data center to connect to. There are some great load balance solutions out there now. One I greatly recommend is from A10 networks. Check them out, there are some real advantages to their solutions. One more comment on the MPLS network option is that you can push your security to the MPLS cloud and have your firewalls, IDS, DLP all sitting in that cloud to protect the entire private MPLS cloud you installed. I would keep your data center to data center sync line direct Point to Point. Just my preference. The main point here is don’t forget the DR portion of your planning. Very key!

The other question was, should I just push everything to the cloud now and be done with it? So given the information they shared, I don’t think you can push your entire data center to the cloud. Things like email, and even your voice services can go to the cloud. But your mission critical systems – can you really get them into a Cloud offering and deliver at or above your current operation? Probably not. What about your corporate data, is the organization comfortable with that data sitting in a multi tenancy environment out of your control? Probably not. So look for the quick wins and easy decisions to make to get some of that out of your data center today. This will help take the load off your aging environmental components while you determine your course of action.

Some of the feedback I get is to just say “if it was you, what would you do….”. I have tried to not do that, but I know folks are interested in my opinion. Given what I know from my colleague, I would build a new data center on premise, that is much smaller than what you have today, and that brings much needed automation, and process improvement. I would place your MDF in that new data center, your key infrastructure components, and mission critical applications. I would turn your email, video conference, voice services, and SharePoint into SaaS solutions. I would also drive virtualization – nothing moves from old data center to new data center without a new plan. New virtualization plans for server, storage, and desktop. I would develop a hybrid cloud solution and look for some appliance solutions for the integration to your external cloud solution. I would look for a storage IaaS solution for your archive data – encrypted of course. I would also build your MPLS WAN for all site connections and put your security in the cloud as a service. Let the security experts do that for you.

This accomplishes a few things, 1) you remove the risk of a facility failure, 2) you take the load off your limited staff and let them focus on mission critical components, 3) you start the cloud journey small and grow into it as it makes sense, 4) you are now in a position to deliver a more successful DR plan to the organization, 5) you will drive down cost(s) with your new facility with the new technology and new approach.

I am not opposed to co-location solutions, I just have found the TCO for that solution hard to sell. You add up all your cost(s), risk(s), risk avoidance, and operational changes, and you can no longer afford it.

Keep it positive!



Scott Arnett
scott.arnett@charter.net

No comments:

Post a Comment