The pair are now very much the primary candidates for supplying open source cloud management services to those enterprises that seek a degree of certainty in what they build their businesses upon. Open source cloud – OpenStack included – has been seen as something specialised, intended for specific areas of application where `hand-crafting’ the management environment is a good option.
But according to Mark Smith, SUSE’s Senior Product Marketing Manager for OpenStack, the options available for using it are now like our relationships with the motor car:
It is like a car. You can buy a kit car and build it yourself. Or you can go to the local dealer and buy one and drive away with it. Alternatively, you can use a taxi service.
The `taxi’ is, of course, a reference to Rackspace and other variants on the public cloud model, and for the latter two of those alternatives, product maturity is an important benchmark. Signs of that with OpenStack include, Smith suggests, SUSE’s shift of the upgrade release cycle from six-monthly to annual.
The company has also added non-disruptive upgrade facilities, so there is now no down-time during the upgrade process. He reckons this maturity means that it now provides the type of stable platform enterprise users want for an enterprise-wide capability.
Not just net-new cloud apps
So this means enterprises can work not just with net-new cloud-native applications, but also evolve their existing datacentre infrastructures to work on the same platform. It can now run containers, virtualisation or bare metal all on the same platform.
Smith ducked the opportunity to give a metric for the number of enterprises now setting about such a transition, which suggests that the majority of its business is still with the enterprises looking to fire up those net-new cloud native applications.
But he indicated the OpenStack community is well aware that the market need now is for a platform capable of hosting the three main options enterprise users are targeting: net-new cloud apps, hosting established legacy applications that can run in cloud environments, and providing an environment capable of helping users adapt business critical applications to run in the cloud.
To this end the SUSE distribution now includes tools to run IBM mainframe applications, a capability that existing enterprise users have been requesting.
So in Smith’s opinion, the SUSE OpenStack distribution is now seen as a `done deal’ in terms of being a mature environment that is now production-ready for whatever workloads are required.
The problem now being faced by users is that, if they have applications running on hardware, at some point they have to change the hardware, and if they are running on OpenStack they don’t have to worry about that, it is abstracted away. That would have been called the inflection point, but they don’t have to worry about it because it is running in the cloud and everything is abstracted to software defined infrastructure running below it.
SUSE provides support to users in the transition process and the Deployment Framework, particularly when the transition includes legacy applications and business processes, which often require bare metal infrastructures and high availability for compute, networking and storage. The production version of SUSE OpenStack provides this level of management and resource allocation automatically.
There is a sticky reality here, however, especially for larger enterprises. Many of them will have long-standing, business critical applications running on technologies from Oracle, SAP and others. It is not the technology per se that is the problem, however, it is the licencing requirements that underpin things.
There are challenges. Every company is moving towards a multi-cloud environment, with public cloud and private clouds, and the question is how you combine those two into something more hybrid. But the question for many customers now is – how do you migrate? One way is to use bare metal solutions. The most economical way of doing it is to maximise the use of a private cloud and use the public cloud as and when you have to.
But the users now have the choice, and it is becoming a business and process argument, not a technology one. That being said, users have to look at their processes, for they will change. If they don’t look they won’t get the efficiency advantages from the cloud. If they still have to have a paper trail to spin up a server there will probably be no advantage for them.
Here comes the 2nd tier, seeking MSPs
Over at Red Hat its OpenStack General Manager, Radhesh Balakrishnan, sees that maturity presenting itself as a growth in a second tier of users emerging, those that want to consume OpenStack but have other priorities, which means they would like it to be managed by someone else.
This led the OpenStack Foundation to announce much greater support for Managed Services Providers at its annual bash back in May, a marketplace that Red Hat has been targeting for a while.
Rackspace is a classic MSP player and recently was awarded the Red Hat Innovator of the Year award for all the work it has done in this area. We now have also got the MetaCloud division of Cisco and IBM Bluemix offering RH OpenStack into the MSP marketplace. I view this as just another consumption model and we are looking to grow it to a network of around 10 well known regional Systems Integrators that know their customers - and their customers’ customers - needs very well.
We have been working with the MSP community for over a year to ensure that there is a flavour of OpenStack that fits their needs in providing the managed services.
Red Hat is also finding that the growing maturity of OpenStack is scratching the enterprise itch of what to do with legacy applications, though he indicated that large enterprise users are still generally treading carefully.
Santander Bank is one large enterprise customer that is already starting down the of determining which workloads are appropriate to move to OpenStack, and is using the company’s discovery workshop program to help it identify and move workloads.
Cisco is also looking to move workloads from VMware to OpenStack. But it is not a case of hit the button and they will all move over. It still has to be very carefully crafted.
In many ways, the aim now is to get enterprise users saying `we need to get away from having to spend all this money again. This one will be expensive but it will be the last time’. As Balakrishnan put it, this is the bridge that has to be built.
If the contention that moving to OpenStack marks the beginning of an end to enterprises fretting about the next (damned expensive) technology inflection point the OpenStack may yet prove to be an invaluable option. I guess we will only really know the truth of that when we see what constitutes the next inflection point, and how well OpenStack copes.