Catalog your way to the Cloud

I have been speaking with allot of customers regarding their own Private, Hybrid and Public Cloud initiatives.  One area of struggle  for customers is articulating the business value of becoming ‘Cloud like’.  The justification tends to diverge towards the ‘ilities’ adjectives; flexibility, scalability and I once heard ‘cloudability’.  But can you really get started with a Cloud strategy if your goal is cloudability?

A proper Cloud strategy begins with some pretty mundane tasks; I  personally think of it in terms of Categorizing, Cataloging and Optimizing.  From an enterprise perspective you need to categorize your workloads so you understand both the attributes and criticality (Whoa, used one of the ‘ilities’).  For example are they Tier 1, 2 or 3 workloads?  Are they Transient or short term workloads, are they infrastructure related? Do they require additional capacity for only certain periods of time  (Bursty workloads).  Once categorized you can begin the next piece of the puzzle, the creation of the Service Catalog.

The development of a Service Catalog  helps you understand what service or services you will be providing, in addition it translates a disparate set of IT resources in a way that makes them understandable to the business.  Categorizing first enables you to determine what Service Catalogs you should begin with.  It allows you to quickly see the real value and understand the nature of the workload.  In addition it can help you intelligently apply costs.

As an example lets take two scenarios: The IT department at company “A” has never given much strategic thought to their virtualization platform and just licenses everything at same feature level, let’s call it enterprise.  When they move to Cloud they calculate the cost of applying the Cloud Management Suite on top of everything.  They struggle to articulate the value and justify the cost so they can’t get the initiative off the ground.

The 2nd IT department at company “B” begins planning for their Cloud strategy by categorizing all their workloads.  They develop a strategy to manage everything through a Cloud Management suite but introduce a mixed hypervisor environment.  This reduces the licensing in DEV\UAT but slightly increases the cost of their virtualization environment for their Tier 1 workloads.  The Cloud management framework will be collocated with their Tier 1 workloads as it will be critical to their long term strategy. 

In addition, long term they want to look at taking their mixed hypervisor environment and move the Transient workload to the public cloud environment.  Because they have chosen wisely they will use their Cloud Management Framework to manage and apply policy across all parts using the same toolset.

Company “B” can clearly articulate the value; they are reducing costs in areas not deemed critical to the business and justifying the tools they need to manage a Private or Federated Cloud environment to get it done .  In addition they are able to identify what business services they will provide as they have reviewed which Service Catalogs they will create in the short-term, near-term and long-term. 

Because company “B”’s Cloud strategy moves forward they are in a better position to optimize the environment to deliver the most benefit to the business moving forward.

Although somewhat simplistic, the point is that the general approach that we took with virtualization does not translate  well to a Cloud Strategy.  While we needed to understand the technical aspects to virtualize workloads, we need to add the understanding of business value to develop our Cloud Strategy.

The allure of Nutanix

The business pedigree of Nutanix derives from the founders who have worked with Google and Oracle Exadata.  The co-founders, Dheeraj Panday and Ajeet Singh set out to create a better platform for virtualization.  Nutanix is on their 3^rd round of venture capital funding which has been successfully awarded.  Rumors of an IPO perhaps explain the ability of Nutanix to draw talent. 

They are very similar to vSAN, although they have the distinction of being first in the market and really view vSAN as validating the approach as well as competition.  They are the fastest growing hardware company in the last 10 years which is interesting as they go to great pains to explain they are a software company.  They actually do not manufacturer hardware themselves but OEM all components.

The platform provides a combined hypervisor and logical storage system that is created from local SSDs and host based hard drives that are aggregated to provide a single volume logical SAN.  They use their own proprietary file system under the covers but present to virtualization farms as NFS.  They do not use RAID but instead allow you to configure a replication factor on the logical volume.  Min is 0 designed for Hadoop clusters however 2 is recommended for virtualization.  This means that you lose ½ of the capacity in a virtualization farm however.  They provide a small form factor which can be bought in certain models designed to run a predetermined number of workloads. 

They have many enterprise class features; dedupe; predictive failure on hardware components as well as a phone home service, dedupe of memory and SSD with disk likely in the future.  They consider themselves enterprise class Converged Infrastructure (CI) although traditional CI vendors would dispute this as they do not apply engineering at the hardware level.  The sweet spot for this technology seems to be VDI although Nutanix targets other applications of the technology such as Hadoop.

Nutanix represents a new breed of CI which I have coined micro-converged infrastructure; a host hypervisor and SAN built from aggregating locally installed SSDs and host hard drives.  The architecture is compelling as it allows a very modular approach at a reduced price point without sacrificing performance.  It is clear from Nutanix’s growth that they are meeting a demand and garnering allot of interest.  For an overview of Nutanix is two minutes have a look at this video.

The Conundrum of Cloud Adoption

It is typical to have not seen much personnel or budget increases in many organizations over the last number of years. User expectations have dramatically increased however because of the consumerization of IT. The new standard in the minds of the users we service is the ‘appstore’ approach to delivering services. In addition most IT teams are still aligned in traditional technology silos and struggle to adopt “as a service” models.

These problem are compounded by where most IT teams spend time. In many recent discussions with our customers at the executive level, they estimate they are spending 60 – 70% of their time on lights on activities, straight operations or firefighting. The remainder of time is spent on new project initiatives leaving little time for proactive strategies to adopt new service models in End User Computing (EUC) and Private or Federated Cloud. How then can they break this cycle and move the blocks forward to become more proactive is often the question asked of us?

The first step is to carve out time that does not exist. The only way to do this is to spend less time in operations by undertaking the following:

1. Categorize, Automate and Orchestrate

You have two choices on where you begin; datacenter or end user services. In the end it does not matter and this tends to be dictated by each customers situation and the current needs of the business

2. Standardize a percentage of IT as a catalog

As much as we want to believe it not everything is unique; IT departments treat everything as custom builds when not all requirements are. Even if this starts as a very simple catalog; office applications (EUC) or a simple two tier VM configuration (Private Cloud) the importance is to work yourself through the process.

3. Turn over delivery

Once you have completed step 1 and 2, enable a suitable business group that is not traditional IT to self-provision. In the initial adoption this may not be end users; it may be business or application analysts; “get it done” as this is when you reduce time in operations. Prepare yourself as this will be a learning process so it may require some heavy lifting in training to the transition team.

Once you have achieved the first three steps you can move to higher value activities described in steps 4 and 5.

4. Adopt a Virtual Datacenter

The vast majority of IT are virtualizing OSes however turning your datacenter into software is a huge enabler. It allows you to thin provision a host of physical devices including networking and storage.

5. Apply Security and Policy to the Virtual Datacenter

Ensure that the virtual datacenter is more secure than your physical one; software can be encrypted and encoded and policy driven. Apply these to your virtual datacenter

After turning your datacenter into software you are ready to drive efficiencies by implementing steps 6 – 8.

6. Evaluate where you are doing your computing

Once your Virtual Datacenter is secure and encrypted it can be migrated anywhere so take advantage of lower cost opportunities to run your IT services

7. Evaluate the efficiency of what you are delivering

Okay we have come a long way however the journey is not complete; evaluate whether it is more efficient for you to deliver each service internally or through a 3^rd party. It is important that at this point it is an evaluation; there is one more critical step to complete before farming a percentage of services.

8. Brand

You are the department who has serviced your users effectively for years; your intellectual property is the knowledge and understanding of the users that you serve. They should always continue to come to you for all requirements even if some are being delivered through 3^rd party. When you farm the services that you can deliver more efficiently elsewhere ensure that the users do not see whitespace between the internal IT team and service providers. Your role in this new model is to be the one stop shop from a provider perspective and maintain quality control.

In these discussions we did not mention Cloud or Federation or any the terms that are wearing thin due to the promise and expectations being out of alignment. We have described a process that enables Cloud adoption. We also did not discern where it gets adopted first; in the datacenter or in end user computing, as this is a business decision. The process starts with an IT department that has no time which is often the situation. It is hoped that in describing the process logically that this will help internal IT team organize their own initiatives and determine what areas to focus on first.