We have attended Microsoft Ignite conference which took place in Orlando in October 2018. As usual, it was a huge event with about 30 thousand attendees. Together with Ignite there is Envision conference which is intended for C-Suite (CTOs, CEOs, …). One of us was also there but this post is just about Ignite.
There are tens of concurrent session running for the whole week in parallel covering broad fields of Microsoft industry e.g. Office365, MS Teams, Azure, AI & Machine learning and developers track. I was personally most interested in Azure and in developers track.
CosmosDB Multimaster
Probably the most exciting announcement for me was the Multi-master support for CosmosDB. We see a big potential in technology especially when we speak about high availability. We can deploy a solution to multiple Azure datacenters (DC). In case of one DC failure the other can handle the traffic.
Our apps are currently running in 3 DCs across the whole globe and we are sometimes struggling with data consistency across different regions. We believe that the Multi-master option will help us to solve this issue.
We plan to pilot Multi-master in Mobile notifications service which provide API for our mobile apps and their mailbox.

Service Fabric Mesh
Service Fabric Mesh is a fully managed service which enables deploying microservice applications without managing virtual machines, storage, or networking. It was put in public preview during the Ignite.
For our microservices, we use Service Fabric right now. With Service Fabric we have quite big overhead for just setting the infrastructure. It is not easy to scale Service Fabric – you have to scale separately virtual machines and the number of application instances. Especially when you compare Service Fabric with Azure App Service it is hard to convince developers to be willing to deploy their apps to Service Fabric. With App Service it is much easier to do nearly everything. It is more a platform service.
Service Fabric Mesh is the next step to platform services. If you want to spin up a container, you just specify the number of CPUs, amount of memory and number of instances and it runs. Additionally, you can specify rules for autoscale. We really like the approach as we can more focus on the apps rather than investigation of some infrastructure failure.
Conclusion
Except for those mentioned topics, there was a big accent to IOT scenarios especially Azure IOT Edge which is bringing some machine learning staff to edge devices which are in the field without connection to the cloud.
The hype around AI and Machine Learning is still growing. It’s being integrated everywhere but from my perspective, it is still more marketing than real usage.
It is worth to mention the official presentation of Kusto engine (a new name is Azure Data Explorer) which is the background for Application Insights. It is possible to feed it with any data. It can ingest a huge amount of data and perform ad-hoc queries in real time. It seems to be expensive compared to Application Insights. It is still in preview so maybe the pricing will be adjusted later on.


Leave a Reply