We are working on a project where we have a need to run our “agent” service in customer’s on-premise infrastructure as a Windows service. This is not a simple utility rather a complex system with about 10 to 12 microservices within it.
Here are general (and very high level) requirements for the service:
- Highly concurrent, fault tolerant, scalable and resilient. Hard to crash and should require no (or minimal) administrative overhead.
- Ability to have workers processing events from a queue.
- Single deployable binary with an embedded storage / database.
- Ability to maintain live & secured connection with its SaaS based central counterpart.
- Integration with various Windows specific tools such as SCCM, AD, PowerShell, etc.
- Auto-updating when new versions are released.
I have been doing a lot of research lately to identify the right tech stack for us.I have narrowed down to two choices: Golang and Elixir.
I have extensive Golang development experience. It is extremely simple to cross compile it to target Windows and generates simple deployable exe in the end. However, in my experience (and opinion), it is well suited for microservice based cloud / SaaS environments. Or a simple purpose-specific OS utility application. Using Golang will lead us to write a lot more boilerplate as well as, a bit of infrastructure to enable hot / on-demand upgrades.
From my research so far, it appears Elixir really excels in all of these fronts. However, I have few concerns mainly due to lack of my knowledge and experience with Elixir.
- How easy would it be to generate a single binary to run as a Windows service?
- What about protection of IP? I know that as soon as any Software gets on an un-trusted hardware, it is prone to be reverse engineered. However, how easy or hard would it be to break into Elixir based system compared to Golang?
I would also be interested to learn about any other gotchas, potential things that can go wrong or I should keep in mind,