How do you provide innovation to your business or customers? Is installing the latest technologies what you deem to be innovation? Is providing cost savings innovation? Do you improve service levels by doing this?
In my humble opinion, it’s a combination of these and more.
Infrastructure practices, be it in a consulting capacity or within an internal IT organisation, provide the foundation for the successful operation of a business that relies on technology to supply goods or services to their clients. They lay the foundation to provide a stable and secure environment that can then be used to provide applications and business workflow to users. I have referred to this for a long time as the ‘Lego block’ model of IT. Whether it be building a green fields infrastructure, or transforming an enterprise IT environment, each component must be built and then added by connecting the blocks and providing integration and integrity to the overall design.
This can be carried through from the data centre, network, storage, server, operating system, authentication, and finally to the application layer. In fact, the applications layer can use this same mentality to componentised services, particularly with web services.
Why is it that most IT departments don’t work like this by default? In most organisations, IT departments are stuck in vendor mode. They select a preferred vendor for a technology, and they stick with it. But the key word in the previous sentence is ‘TECHNOLOGY’. I am not averse to using a single vendor to provide a solution; in fact there are some great synergies in doing that. But my concern is that IT shops (in particular, infrastructure) will purchase the technology because of the features, bells and whistles, and not because of the ability to develop a service around the technology.
Let’s take virtualisation as an example. Forgetting the mainframe days of the 80′s where CPU time was purchased by the minute, the hypervisor (thin layer of software that allows abstraction to host multiple systems on a single server) was first placed onto a desktop system and allowed the uber geeks out there to run two copies of Windows 2000 and get their paper MSCE (I know, very tongue in cheek).
Fast forward to 2009, and not only does server virtualisation software allow you host hundreds to many thousands of ‘virtual guests’ across minimal server hardware, it has the framework and management capabilities that we used to use 3-5 separate products to achieve the same result. The virtualisation vendors have developed, or acquired, the management tools and integrated them into the mainstream product offering to allow IT organisations to provide a service just by buying their software. If we take this back to the building block mentality, the virtualisation vendors is the single block. VMware, Citrix, Microsoft or Redhat, it’s a holistic service you are purchasing.
So what do I think constitutes the building blocks of a great infrastructure environment? I’ll start from top to bottom, but read up to understand the layers:-
Infrastructure Services
- Application Platforms (web services/databases/multi-tier) – Controversial I know, but applications teams shouldn’t own the platforms that they develop on. When it comes to providing a service level, applications can be the most difficult to monitor and measure. App Platforms might make up multiple services, but all the same, the developers should be requesting access to components such as a database, a web service and an authentication service and providing the workflow and access methods only and allowing the service bus or user to populate the data. Bring this into the infrastructure team and put process and supportability wrappers around the platforms.
- Authentication Services – If you can’t provide a secure service, you won’t be able to provide a stable service. Have one core identity source, and use this as the source of truth. Traditionally this would fall under the operating system building block, but yank it out into its own service. Know how much this cost to run. Automate it and allow self service management. Measure the service management practices and costs. It helps to justify head count later when HR wants to know what it costs to provision and manage users.
- System Management – Find a system management platform that can monitor your data centre and preceeding upward building blocks holistically. Make sure it integrates with the platforms, networks and storage layers natively. If it doesn’t, enquire into the third party integration tools. If you have a small environment, don’t go out and buy the Mercedes Benz model. Think Open Source and get a consultant to install it. Some of the Open Source products are MORE than capable.
- Operating Systems – Windows, Unix/Linux or Mac? Does it really matter with enterprise software? Unless you have a distinct requirement to have a mixed operating system environment, stick with the core knowledge of your support teams, and keep it simple. Remember, that as a service, you want a locked down, secure known environment that you can build a rock solid service level with, and make changes easily.
- Virtualisation/Platform – As I have already discussed, you are buying not just virtualisation, you are buying a management framework. Choose the platform that you can identify with, and scale out as required. Make sure you can measure and monitor it and even if you don’t intend on using chargeback to the business, understand the costs of all the options so that you can see that you are providing a good value service to the business.
- Servers/Storage – I must say I see this as the largest commodity within IT infrastructure. There are hundreds of solutions out there, and most provide the same functionality at different costs (I’m not excited by the latest developments in storage). Think carefully before signing the multi-million dollar lease deal and beware of vendor lock-in. You need to provide an open storage system to the layers above to provide a true storage service. Power comes in 2 flavours, AC & DC, and you can covert both ways, why would you want to have three or four flavours of storage that you can’t move between?
- Network – These are the plumbing pipes that allow your organisation to provide a stable IT environment. I include storage networks in here. Redundancy and bandwidth is key. Capacity planning is essential when planning your network. In the data centre, the port in the rack is the service. The internet feed, the DMZ, the port on the users desk is the service. Know what the monthly cost of providing this is.
- Data Centre – You take the utilities and integrate them into delivery of providing power to your cooling solution (Free Cooling I hope), racks, batteries, etc…The data centre itself is made up of service offerings. Each deliverable is measured, monitored so that you know the true cost of providing the service to your business.
- Utilities – These are the services supplied to you by the Electricity, Gas, Telecommunications, and Water companies. Each of these is provided with a cost and a service level. Hopefully you have redundancy for each of these services and it’s built into your SLA? Are these now sourced from a green power provider?
…of course all of these building blocks should be developed in tune with a service management and security framework. As these service offerings mature and you move on from provisioning and supply then, only then, do you look at service improvements and incremental updates to the service offerings. Integration points between each of these blocks are key to providing a mature service to your business.
If you can find the right architects to guide the business on how to better utilise what technology you have already invested in, the more cost effective and efficient your infrastructure organisation will be.
