Why have a strategy for cloud management?

Source – cloudcomputing-news.net

With the general cloud hype and the widely declared corporate intent to base future business solutions on cloud technology stacks, businesses may take the view that deploying cloud solutions involves little more than accessing a vendor’s cloud portal and following a few clicks, have your virtual services deployed and ready for operations.

Whilst this approach will give an enterprise a cloud-based virtual environment, what’s delivered is far from being part of a corporate information ecosystem.

Businesses need to view cloud as providing a virtual data centre, in which they store virtualised versions of traditional infrastructure, including computing platforms, storage, operating systems, networks, firewalls, gateways and Internet/private connectivity, all of which needs careful design.

The major benefits which businesses will see are an OPEX-based cost model with vendors actively competing on price, fast times to deploy, and the flexibility to dynamically right-size environments.

One of the often overlooked challenges of deploying solutions into this new environment is around how to monitor and manage what is essentially a fairly dynamic set of services. How to maintain an accurate record of the components which have been deployed when there are specified policies to auto-scale these environments on demand for example?

Responding to an anomalous event or incident from a cloud solution is particularly difficult if your IT service management solution has no record of the virtual devices which have generated it. Also, in an increasingly complex cyber-landscape, how can businesses detect the difference between normal workload scaling throughout the day or in-line with customer demand, and a DDoS attack, in order to respond to each appropriately?

At risk of complicating matters further, most large organisations will likely be deploying solutions to multiple clouds from multiple providers, integrating these with in-house legacy estate. I also expect compute loads to be continually migrated between cloud providers, as each offers increasingly compelling services and price points, so far from offering a simplified IT environment, it’s clear that cloud adoption will drive a new set of service management challenges.

Traditionally, many organisations have relatively static IT estates, and have incrementally procured and integrated the service tooling to monitor and manage these estates, often from multiple vendors, with the on-going cost and risk of this tooling integration lying with the end-user organisations.

Whilst a degree of syntactic integration is typically achieved between tools via this approach, deep semantic integration generally is not. Consequently, it’s becoming clear to a number of public and private sector organisations that their existing service management tooling won’t be able to effectively manage cloud-based solutions without a tightly integrated set of tools based around a single, consistent view of the estate under management.

Evidence of this growing challenge is found in the growing number of cloud management offers in the market, where this complexity can be managed for you at a price, with Gartner for example now publishing assessments of the capabilities being offered. These providers include established IT consultancies, traditional hosting providers transitioning their business to cloud management and a number of new challengers.

Looking across the market however, most large organisations are currently opting to manage their own cloud or hybrid estates and will almost certainly encounter the various challenges which I have described above. So, in order to ensure the cohesiveness of the cloud ecosystem of an enterprise, the following learning points should be built into cloud management strategy:

  • Ensure that the cloud management solution is based around a single view of the inventory under management, both real and virtual, with all tools sharing that view. Using separate inventory views for different tools will create inconsistencies, degrade the overall functionality offered and present long-term integration challenges.
  • Deploy a robust capability to discover assets deployed in both cloud and legacy environments and their respective dependencies, using this to keep the system inventory current, tracking demand driven, automated cloud deployments for example.
  • Consider deploying tools which offer policy-based automated remediation capabilities in order to drive down cost of service, and provide timely resolution of incidents.
  • To provide consistent deployment patterns across different cloud platforms in-line with internal policy, consider deploying an enterprise class cloud management tool supporting policy-based deployment patterns, alongside an overall governance framework. This is especially relevant where third parties or contract staff are provisioning cloud solutions on your behalf, and their compliance with corporate policy needs to be managed.
  • Look at options to converge cyber and environment monitoring and management, in order that both views of the estate under management can be reconciled to a single solution view, allowing automated remediation capabilities within the service management toolset to deliver policy-based interventions in detected cyber incidents.
  • To avoid piecemeal tooling integrations and related on-going costs, look at solution stacks from a single vendor, or alternatively options where a single vendor will underwrite the on-going integration of third party tools into their stack. Also, use tooling configurations which are as close as possible to the vendors’ defaults, avoiding excessive configurations which will drive up both deployment times and lifetime ownership costs.

In conclusion, whereas cloud offers significant potential to increase business agility, whilst driving down the investment required to achieve this, businesses must not neglect the complexity of managing cloud ecosystems. A key test for any business case for cloud adoption within an enterprise should be whether in includes adequate provision for end-to-end management of the solution, as without this, the probability of the described benefits actually being realised reduces significantly.

Leave a Reply