Going beyond SDDC and Hyper-convergence!
Enterprise IT has been going through a lot of changes over the last decade given the shift of IT from virtualized environments to a private or public cloud, containers, new orchestration layers, new application description formats and the conversion of appliance-based service delivery to virtual appliances running on your infrastructure.
One of the common themes across a lot of these new technologies has been fast evolution, rapid iteration and the need to integrate them together. This puts the burden on the IT staff to keep up with all that is happening, while doing their day-to-day job. So all the agility benefits from cloud, containers and automation come at a cost of increasing complexity. Here is a curve that best represents what a lot of customers and IT teams are feeling:
The question is how to tame this complexity in a datacenter?
A datacenter typically has anywhere from hundreds to thousands of devices. On top of this infrastructure, customers typically install four kinds of software:
– Management software for the infrastructure components
– Management software to convert the infrastructure to a cloud
– The business applications
– Monitoring and operations software to manage all of the above
This requires a team with a lot of expertise to deploy all this hardware and software.
There is a precedent where a similar kind of complexity was solved by completely eliminating all of the on-premises components: Salesforce! It converted the CRM application to a pure SaaS-based consumption model. This is exactly what is done by public clouds for infrastructure to simplify the operations and management complexity. Public clouds essentially provide just an API endpoint to talk to a pre-installed cloud. The customer just has to deal with #3 in the list above and run the monitoring tools for their application.
Salesforce was just an application, and the growth in Salesforce consumption by an enterprise is directly related to that company’s customer base. Hence, the growth is much more organic and the cost grows in a very sub-linear manner. In the case of a public cloud, the enterprises are consuming actual compute, storage and networking resources; these do not scale well in terms of cost. In fact, the cost increases much more rapidly and very soon an on-premise installation starts to look cheaper, if somehow the complexity and Opex can be minimized. Secondly, the growth of consumption may not be directly related to number of customers. Lot of workloads are run just to do software development, provide ‘freemium’ offerings and these costs directly impact the bottom line for the enterprise.
So there is a need for a solution that can tame all the complexity while running workloads on-premise on a company’s servers, storage and network. In many cases, there are other reasons to do so, including data locality, performance, security, visibility and control.
The real question is how to provide a public cloud-like service with on-premises servers and storage without adding a huge amount of operational complexity.
The answer is: Cloud Managed Datacenters
Cloud management is an innovative approach that leverages a SaaS platform to simplify installation, monitoring and management of remote datacenters. The architecture and rationale is very similar to what Meraki did for networking.
In order for this architecture to be beneficial, the infrastructure deployed in the datacenter has to work well with the cloud management built into the SaaS portal. It must be an integrated solution combining a set of server and storage devices and a SaaS-based web portal designed to operate together.
Some of the key benefits of cloud managed datacenter (CMD) include:
– All patching and upgrades are driven from the cloud over a secure connection
– No manual software installation on-premises
– The existing IT staff doesn’t have to understand all the details of cloud software
– The SaaS layer has high feature velocity
– Part of IT operations can be offloaded to the CMD vendor
At ZeroStack, we have built a similar solution for datacenters. The solution consists of two parts: datacenter infrastructure and a SaaS based web portal. Once the servers are installed in the datacenter, the rest of installation is driven by the SaaS portal. One can simply create an account on the SaaS portal, get an account code, enter that into servers and start managing them as part of an on-premise cloud. All the management, monitoring and consumption is done using the SaaS portal.
In addition, the solution is integrated with public cloud via the SaaS player and web-based GUI. This allows the SaaS platform to deliver features at the same velocity as the public cloud vendors and keep up with their feature agility. This is something that cannot be achieved by vendors providing on-premise, shrink-wrapped software that is released once a year.
We believe that this is the only model that can simplify the complexity of building a cloud and delivering a private or hybrid cloud to customers.
Lets bend the complexity curve using Cloud Managed Datacenter!