Many CTOs are telling me that they do not use the term Cloud anymore. Instead the term "utility" seems to represent better what capabilities they are after – finding a way to release the “legacy”, “out of support”, “silo’ed” shackles by making use of new technologies that can consume services from legacy, on-premise and / or off premise (public).
Every time I engage with a large organisation the expectation is that there are many large / very large “to-the-cloud” transformation projects and that we “simply” apply the templates and case studies to acceleration the transformation. But the actual deployments are low.
As per Gartner’s prediction, “nearly half of large enterprises will have hybrid cloud deployments by the end of 2017, Hybrid cloud computing is at the same place today that private cloud was three years ago; as actual deployments are low, but aspirations are high”. Other analysis suggests that that roughly 5% of all compute sits in / on a “cloud-based” platform. So, still a long way to go.
So what are the main challenges when starting on a hybrid cloud adoption:
Driven my market hype there is an expectation that moving “to the cloud” Enterprises can save cost. However, this is not always the case. In environments where the asset is >5yrs old, where the Data Centre facilities are covered in a separate (to IT) budget it will be hard to reduce opex.
On-premise traditional IT:
Successful hybrid cloud requires well-architected private cloud. Traditional IT have to transform to a private cloud to include self-provisioning, dynamic resource scaling, a chargeback model for lines of business, orchestration for automating repeatable tasks and a high-visibility management platform to monitor how and where services get deployed.
Cloud is not = Cloud:
You need to treat your cloud provider selection like a COTS selection. Each cloud provider will have very different characteristics covering SLA, APIs, Monitoring, Nr of DC, Instance Types, Supported Operating Systems, Standards, Ease of migration, Standards, Service Age, etc. Make sure you assessed all characteristics to ensure this is the right platform.
ensure you are covering all aspects from user identification, authorisation and authentication, to data and network encryption, intrusion detection as well as clear separation with co-hosted services.
latency and bandwidth issues, chatty applications, network topology, network security, extending local topology to the cloud.
Data and Application Integration:
throttling of inbound queries, data move, interdependencies and impact of remote host is not necessary available.
Service, System, Release & Change Management:
configuration management, patching, capacity planning and monitoring, system monitoring etc.
moving machines, metadata and configuration.
cultural shift for people, skills are in high demand so difficult to find, require architecture skills.
This can be quite tricky as seldom you have a 1:1 map between infrastructure components (like server) to application. Many application can "live" on multiple servers, having near real-time interfaces with other applications etc. It is sometimes a bit like trying to pull one string of spaghetti from a large bowl of spaghetti bolognaise; meaning you have to touch more than you really want. Add the lack of 100% accurate As-Is information plus the fact that the environment is seldom fixed, it can be complex to gain a clear view on current, transformation and future cost.
So what is the way forward then? If cost saving is your target then you should consider a number of options.
- Defer noncritical key initiatives.
- Re-examine networking costs.
- Consolidate Apps and Infra.
- Virtualize compute, storage and/or network.
- Make use of new utility / cloud based technologies (private and/or public).
- Reduce power and cooling needs.
- Contain storage growth.
- Deploy “Self-Service”.
- Streamline IT operations.
- Enhance IT asset management
- Optimize multisourcing.