MikeDoesCode

Arduino and C++ to work on the ESP range of embedded wifi chipsets but has also let me get into working with other dev boards like cheap allwinner tech stuff for fun for now.

Docker (finally) in a production setting rather than just messing around a bit with it, finally taking the time to work through k8s, swarm and service fabric for differing orchestration methods on different platforms.

Graph databases, like... most of them now, I will be continuing to work and play with Dgraph, but Neo4j and all of the tinkerpop stuff is pretty great if still a little lacking in docs for some of the implementations.

Despite the bulk of my professional development being on .Net applications, this has also been the year I moved my primary workstation from Windows to Linux and I haven't looked back almost at all but I have booted my old windows machine up as a VM a few times when I ran into the occasional brick wall, but that's become a lot less frequent as I've gotten more comfortable with Fedora.

They have "bounded staleness" as an option for consistency, but obviously it doesn't break the laws of physics, latency is latency and you still have to account for that somewhat but at least you have the bounds to work with.

Documentation is still a bit all over the place, at least for the .Net space. Some information is only apparent after diving into GitHub issues, it's not a total disaster but could definitely be better. Especially when some documentation recommends entirely incorrect dependencies or misleading support for some gremlin commands.

However the product itself is great, I've been using the gremlin interface almost exclusively. I will say however, that having used JanusGraph and other TinkerPop compatible products some of the more advanced gremlin features like branch and choose are absent which can make porting some applications problematic.

A further potential annoyance or benefit is the fixed usage of GraphSON for the documents in Cosmos which can make the results of your query appear radically different to what you would expect if you're familiar with other TinkerPop graphs.

Didn't HP design and I think prototype something like this with memristors, calling it 'The Machine'?

edit So HP built one with 160Tb of memory, I remember it being proposed with memristors but haven't been able to check if the prototype used them... Does anyone know what is different about IBM's that let's them claim this as a first though?

When I was working with AliCloud I ran into an issue in that during peak hours, we'd want to scale-up, and they'd be "out of stock" of virtual instances... Which is fine if you have the budget to keep a load of instances running, but if your spike goes over what you expected, there's no resource left for you to scale up. Not sure if that's still the case, but scalability is perhaps the biggest draw for me to the cloud and it seemed AliCloud didn't really get that right.