Infrastructure
Infrastructure is the concrete physical foundation of any IT service. Don’t be fooled by the word “cloud.” Behind every cloud is a data centre with servers, storage devices and network gear. In the past we would take clients around data centres and show off shiny boxes and flashing LEDs. A lot of hardware vendors even made style design decisions about how sexy their product looked. Today you are less likely to walk around a data centre. Google and Amazon are good examples of cloud providers who would rather not discuss their infrastructure or data centres, though they have invested hundreds of millions of dollars to provide a global data centre footprint. So easy, right? Build one big data centre (two if you want redundancy), put all your applications and data into it, and job done.
Unfortunately, it’s not that easy; a combination of data regulations, regional restrictions and speed of access are some of the key considerations. This is why you see the cloud providers standing up more cloud data centres across the globe to handle these requirements. Your business may well be in the same position, and therefore you will end up with a dispersed infrastructure footprint.
Acknowledging that we need good infrastructure, what are the key considerations overall? Is it about the success of securely leveraging the resources you have? Let’s start with the data centre itself. This is a major investment, and running and maintaining these facilities must be considered. Data centres are key resources to be leveraged. Using physical segregated infrastructure within the facility can provide added security, ensuring there is no chance of data bleed between applications, business units or clients. If you have disaster recovery services, you typically need another data centre, suitably connected and managed.
Much focus in the data centre is on network connectivity. In today’s connected world, data no longer needs to flow just within the traditional Intranet networks. In fact, Intranet is becoming a thing of the past. Now we are simply connected; we need to connect to applications and data sources from both internal and external locations. The network should support secure and resilient connections with integrated secure VPNs, firewalls, intrusion detection systems and high availability configurations to ensure services are available in the event of an issue or outage.
In terms of compute and storage solutions, there has been a move from traditional server and storage infrastructure — a “build-it-yourself” mind-set — to the new converged infrastructure, which has pre-packaged server, storage and network products in an integrated stack with predefined sizing and growth options. This can be an accelerator, as these converged infrastructures are pretty much ready to go and can be deployed like an appliance, versus the traditional months of negotiating with vendors, arguing the values of preferred products, and delay in knitting all this together in the data centre. So with converged infrastructures, job done.
Well, not quite. The issue with converged infrastructures is that they come at a price. Typically the products used are enterprise grade, designed not to fail, and have the support and backing of major vendors. Today, these elements are being challenged by the application space, with new apps that are self-healing and able to operate at web scale. Therefore, all the resilience and high-end products in the Infrastructure layer are just adding cost to the architecture. Hadoop is a prime example of an open source product that is designed to be built on commodity hardware products; if a server fails, you throw it away and put in a new one. The cluster reforms and off you go again. As we look at email and other business applications, there are more of these cluster-based solutions that are challenging the infrastructure to keep up and meet the cost-to-function needs of the business.
If we also consider that not all applications need to be hosted in your own data centre, you start getting into hybrid solutions. Although you may wish to host your critical production applications and data in your controlled facilities, this typically means that those parts of your business get caught up in the change control and restrictions typically imposed in the data centre. However, less sensitive environments like development and testing can be hosted outside your facilities. If you use cloud-based services, you can typically improve response times and decrease your time to market.
Link to Part 6 Other Considerations and Conclusion