Many things in life seem simple until we become aware of the mechanics behind the scene. Getting a glass of water is “simple” until we ask questions like ‘how did the water get here?’ and ‘is it safe to drink?’ The cloud is no different, running a VM on somebody else’s equipment sounds “simple”, but this simplicity quickly evaporates when we start talking about VM profiles, managed versus unmanaged storage, and performance limits. However, this complexity and confusion actually begins to dissipate when we start asking questions like ‘how did we get here?’ and ‘how have we been doing it?’.
1. Figuring out Storage
When discussing storage we may hear terms like ‘tiered storage’ or ‘hot vs. cold data’ analysis. This is the process of assigning different categories of data to different types of storage media to drive down costs. IT administrators need to understand what type of data is being stored, how frequently it is being accessed, and the relative importance of the applications it is feeding.
So what is tiered analysis?
Just like cloud providers perform analysis to find where CPU performance intersects latency, storage tiering analysis effectively ranks data and provides administrators with the information they need to place the most frequently accessed data on the highest performing storage (Premium) versus data that is rarely accessed or requires a lower levels of performance (Standard).
Consider the General Data Protection Regulation (GDPR) which becomes enforceable May 25, 2018. Organizations impacted by GDPR will need to understand how personal data is collected and where it is STORED. They will also need to set in place a Data Governance program to manage this data and support the rights of the data subjects underneath it. An initial thought may be to head to the Human Resources department to get this process started, but detailed records regarding what is being collected and where it is being stored may already exist in the IT department as part of their storage tiering analysis. If these types of records already exist then you’re off to a flying start.
2. Multiple types of Storage
Organizations typically leverage multiple types of storage (traditional hard disk drives, solid state drives, and in-memory cache) across multiple vendors. They then need to understand performance characteristics (IOPS and throughput) specific to each implementation, location and the quality of the connectivity, perform regular system updates, replace failed or failing drives, monitor warranties, capitalize then depreciate costs, and if operations need to be re-located, budget for the downtime as storage is one of the last things you can move, but it’s one of the first things you need to bring systems get back online.
In the cloud these issues no longer exist. Cloud providers publish the performance levels they will provide, can provision storage wherever it is needed and are responsible for maintenance and can provide additional storage on demand, i.e. no need to go through additional capital expenditures as we run out of storage or our needs exceed the performance levels we can achieve from our existing equipment.
3. Cloud v. Physical Hosts
Medium to large sized organizations rarely leverage one type of hypervisor, and even if they did, it would be even rarer to see every host running the same version. There are many drivers behind this hypervisor divergence, for example:
- One hypervisor may offer advanced features at a higher cost to support mission critical applications, while others may not offer the same feature set, but are significantly cheaper and satisfy the ‘good enough’ test.
- Mergers and acquisitions are never put on hold or avoided because the two organizations run on different hypervisors.
- Employee background, everybody has their own ‘favorite’ platform, when turnover occurs alternative platforms are recommended over others that may already be in place.
- Upgrading hosts isn’t easy, all VMs need to be relocated to permit the upgrade, and if resources are limited, some systems may need to be shut down while the upgrade takes place.
4. All Hypervisors are the Same
No! This divergence in platforms and versions has created multiple headaches for administrators, for example:
There is no single pane of glass administration: There are tools on the market that can be used to administer multiple types of hypervisors, but the features are typically limited and often result in the administrator going to the platform to perform tasks that go beyond simply powering a VM on or off.
Compatibility: There are multiple tools on the market to convert VMs from one platform to another, but the key element here is conversion. Hypervisors are not compatible by default, and while moving a VM from one type of hypervisor to another may not sound problematic, this requires knowledge of both platforms, downtime, storage migration and finally post conversion testing.
Backup and Disaster Recovery: Few people like planning for disasters, but for those who do, they want a plan that is both easy to follow and will return the organization to full operational capacity as quickly as possible. Having multiple hypervisors, or even if we are able to standardize on just one vendor, having multiple editions, complicates recovery plans and typically places reliance on a larger number of people, if and when a disaster occurs.
5. You tell us that there are multiple hypervisor versions out there, but what proof do you have?
Let’s look at the last ~30,000 hypervisors Movere has inventoried. This is a breakout of the two most popular hypervisors on the market and while most are newer, close to 2/3rds are one MAJOR release off current:
So, has the cloud made all of this more complex and confusing?
It hasn’t! In the cloud we’re renting computing capacity. It’s the cloud providers’ responsibility to manage the underlying physical resources, to manage compatibility, to relocate resources to alternative hosts to perform upgrades and apply patches, to formulate a backup and disaster recovery plan specific to physical resources, and unlike an outage that takes down our only datacenter, no matter how small we are as an organization, as a branch, or as an application service center, the cloud provides us with redundancy opportunities that we simply can’t afford on our own while at the same time eliminating a daunting number of administrative responsibilities.