Assuring the network from the public cloud

What are the benefits of placing operational systems software in the public cloud - and can anyone do it?

Telecoms software vendors are beginning to prove out the public cloud as a suitable location for operational, support and core network software. Athonet has enabled its core network to be available on Amazon Marketplace as BubbleCloud. Affirmed Networks too has a vEPC for public cloud deployment. Optiva offers its BSS software on Google Cloud, and warns that telcos that ignore the rise of the public cloud could miss out on the benefits.

One of the strongest launches has come from OSS and service assurance software vendor MycomOSI. The company, recently acquired by PE house Inflexion with a view to accelerating its cloud-native play, has launched its service assurance suite as a set of microservices available for public cloud deployments. Three UK is its flagship customer, given it is willing to be public, but there are clearly others interested in deploying service assurance on a SaaS model from the public cloud.

Head of Marketing Ian Meakin said that the benefits are many, but here’s a few he touched on:

The first come under the time to market heading. First you bypass the procurement cycle within telcos for the underlying hardware and data centre tech to support the software deployment. “If someone buys our software today it can take three to six months just to get the hardware infrastructure available. We had a real customer take in excess of 18 months to get infrastructure through procurement and so on. With the public cloud version we can turn it on and make it available in less than one hour.”

Secondly, you are not dependent on other bottlenecks being unblocked. “We [service assurance] are at a point in the project where we are dependent on previous things. As an example, if someone is doing a telco cloud stack and NFV, SDN, all that – before we can assure it all that needs to be set up. So to have it instantly turn on is a massive advantage.”

The second set of benefits are all “normal” benefits of SaaS but these are amplified for an operator. For example, true resilience, using active-active mode which means that mirrored deployments are simultaneously active, rather than a back-up being invoked in case of failure or degradation. “If the network degrades and you lose sight of it and are unable to repair it… that makes our stuff mission critical from that point of view. So our system has to be highly available. And to do that means starting to replicate data centres, virtual machines, etc. That just adds cost, time and complexity to what telcos are doing. But with the assurance cloud that is built in under SLAs. We have 4x9s availability; actually we will go beyond that and will announce that very soon. But people only asking for 4x9s at the moment.”

The final operational benefit is in elasticity of the app and the resources to support it.

“So AWS does at some point have finite resources available, but in the context of us and our applications everything is perceived infinite resources. If you need more compute for a machine learning algorithm, more storage and so on, it’s elastic as part of that standard SaaS offering.”

From our point of view we are saying it’s the world’s first carrier grade service assurance run from the public cloud – and nobody is saying it’s not.

So how does Mycom build this, has it been a long road to architect its software for a public cloud deployment?

“From our point of view we are saying it’s the world’s first carrier grade service assurance run from the public cloud – and nobody is saying it’s not.

“So in the world of service assurance, doing that network wide and all the sophisticated stuff, it feels as though we are path-finding a little bit.”

Meakin said, “We started a while back. We chose a point in time several years ago where we said, ‘From this point in time everything will be written in micro-services and for containers’ – that’s the first step. And we needed to implement devops and get into CI/CD thing to support that whole architectural approach. Then we re-architect the existing product set to be containerised, and assess the level of granularity we need – not everything is this small everywhere, there is an optimal level of granularity of micro services.

“But effectively all our stuff runs on a Red Hat container platform, using open source APIs and technologies such as CEPH and Kafka, to provide a layer that makes you agnostic to the underlying infrastructure – and that’s the cloud native part of it.

“It took a while to do, but making the to jump to public cloud was then actually pretty straightforward, although the public cloud environment gives you a lot more things to play with. So with the different operations and management capabilities we see the AWS platform as almost programmable by the application in real time. If the application needs more resources the software will ask the infrastructure to do what it needs to do.”

Of course, the cloud software is not used only to assure virtualised functions and the services that run over them. It must also assure legacy and hybrid networks. “Three UK is assuring its RAN, transmission, backhaul and core and IT systems, everything they run, all domains and vendors from a public cloud service assurance system. It’s the aame system taking data in from all those elements.”

Mycom OSI has worked with Red Hat to provide the assurance part of Red Hat’s reference platform for the virtual Central Office. That’s a a mini NFV-SDN telco cloud stack that integrates tech from a lot of partners.

“We take data off it and assure all the components. In fact we add value by correlating across layers, the physical infra, the VIM, the VNFs, the service chains, modelling that, assuring it and correlating across it. But that vCO stack, as an edge computing thing, is part of the end to end network, and we also do that end to end bit as well as the virtualisation stack and the physical infrastructure as well.”

“In fact monitoring it and making sure it doesn’t break is sort of service assurance 1.01. More sophisticated use cases are emerging for closed loop automation, for self healing, predicting degradations and using performance metrics to see where things are going. That’s been a very interesting conversation to have with Red Hat.”