Cloud Management 2013 Predictions
2012 was an exciting year for VMware’s management portfolio. Some of the highlights included VMworld Europe, where we launched new versions of our management products. In addition, we welcomed DynamicOps and Pattern Insight into the VMware family.
With the acquisition of DynamicOps, VMware wanted to provide its customers with support for heterogeneous infrastructures. You can read more about the work we’re doing in this area in my previous blog. vCenter Operations (vC Ops), vCloud Automation Center (vCAC, formerly DynamicOps), and vFabric Application Director (AppD) all have support for both VMware and non-VMware virtual infrastructure and clouds. vC Ops and vCAC also support physical machines. This is a critical capability as customers have some yet-to-be-virtualized physical machines, developers are jumping out to public clouds, acquisitions are bringing in different virtualization technologies and customers are seeking management products that can seamlessly manage across all these different environments.
We also saw our management team make a lot of strides on the automation front in 2012 – VMware added automation into its entire management product portfolio. vC Ops now has vCenter Orchestrator (vCO) integration, and what’s cool about vCO is that it allows you to design workflow processes visually:
So when vC Ops generates an alert, it can automatically trigger a vCO workflow to create a ticket in your ticketing system or take an action like right-sizing a VM. vCO gives you a ton of options and has plug-ins for popular third-party systems, such as ServiceNow. Next, vCAC enables automation for processes that are largely manual today. It enables you to create a catalog of pre-configured services and allows users to select services to provision and do manage their configuration. Users can select these services and they are automatically provisioned based on the policies you’ve set. Thus there’s no need for an IT admin to manually power-on new VMs anymore. Finally, AppD does for applications what vCAC does for infrastructure. With AppD you can create “application blueprints” that contain full specifications for application deployment. You can even change parameters like environment variables from the AppD UI. This dramatically simplifies application deployment using data-driven automation.
We were also able to make our management products more cloud-aware in 2012. vC Ops has a number of different cloud adapters, including vCloud Director (vCD) and Amazon EC2. vC Ops can provide a single dashboard for viewing performance or compliance across all your clouds, both private and public.
Example custom dashboard for monitoring multiple clouds
This is very important as more developers are bypassing IT and going straight to Amazon. The ability for IT to monitor those VMs and ensure corporate and regulatory compliance is paramount. AppD also added Amazon EC2 support. Again, companies are starting to use Amazon for dev/test and so it makes sense to automate that deployment. (Especially since IT now can monitor those VMs for performance and compliance!) vCAC already had support for Amazon and added support for vCD.
Those were some of the highlights we saw in 2012 – so what does 2013 hold for customers? Let me look into my crystal ball and share my predictions…
Prediction #1: I think we’ll see the convergence of cloud infrastructure and application monitoring. As I discussed in a previous blog post, cloud management is moving away from the tightly integrated vertical silos into horizontal layers. However, that doesn’t mean that those horizontal layers can’t interact. VMware’s first goal was to provide a “single pane of glass” for each layer, which we’ve done with tools like vC Ops, vFabric Application Performance Manager, etc. Now, our next task is to be able to cross-reference application information with infrastructure information. Bringing together these two sets of data will allow for deeper understanding of problems and quicker MTTI and MTTR.
You may be asking, “but how is this different than the vertical silos we started with?” The answer is that the vertical silos were very specific to the environment they were monitoring. With cloud management, what we’ve generalized our approach to management at each level. So the infrastructure layer doesn’t care whether it’s monitoring a private vCloud deployment, Amazon EC2, or a physical machine. Similarly, the application layer doesn’t care what type of application it’s managing. We can then cross-reference those general solutions for converged infrastructure and application monitoring. This solution lets customers manage any application on any infrastructure, which is very different from the tightly integrated vertical silo, which managed only one type of application on one type of infrastructure. It’ll be very exciting to see!
Prediction #2: I expect to see massive scale in 2013. Clouds are continuing to get bigger and bigger, and many customers are running tens of thousands of VMs in their clouds. Some will likely reach hundreds of thousands of VMs next year. This means that management tools must scale to match those numbers. Our team has worked hard to provide the “single pane of glass” everyone wants, but we’re at risk of losing that if management tools aren’t able to meet the scale requirements of modern clouds. If the management tool can’t scale, it will mean that IT will have to provision multiple instances of the tool and there won’t be a single UI or API that ties them all together.
I believe the technical solution to this problem will not be to simply make the single management tool server bigger (scale up), but instead it will be to scale out, either through distributed databases like Cassandra or in-memory caches like VMware’s own vFabric GemFire. In this model, there would be multiple instances of the management tool, but they would talk to each other and federate the set of objects they’re managing. This way, a user could log into any of the management tool instances and see data across all of the instances. However it’s implemented, I see greater levels of scale as a big trend heading into 2013 and a crucial problem for management tools to solve.
Prediction #3: I’ll finish with a bold claim: I believe 2013 will be ITBM’s breakout year. What is ITBM you ask? Well, if I’m correct, by the end of next year you’ll know the answer! (And maybe you’ll even have given it a spin.) In any case, ITBM at its core is about running IT more like a business. While it doesn’t mean you’ll want to aspire to the scale of, say, Amazon AWS, you will want to better understand your costs, benchmark those costs against your peers and the industry, manage SLAs, and the like.
One of the goals of cloud computing is to provide self-service for corporate users and IT-as-a-service in general. This means that IT will no longer be a cost center for a company, but will instead have to pay for itself through the services it “sells” to its corporate users. Thus IT really will have to be its own business, understanding its costs and charging back to the business lines appropriately, all while maintaining SLAs and other service qualities that any “paying” customer of a business would expect. Most of our customers have yet to think about ITBM as they are still wrestling with the basic technology issues of moving to a cloud model. But as they get those technology issues under control, they will need to face the organizational challenges that follow. And when they do, they will need some sort of business management like ITBM.
2012 was definitely a fun ride, and I expect 2013 will be even better! What do you think the big trends will be in 2013 on the management front?