What is our primary use case?
My primary use case is for SOC automation but it's used for a lot more than that. Some of the use cases are more or less appropriate for it. It's capable of doing a lot of things.
We use the SOAR platform to ingest alerts and escalations that we get. They do the actual enrichment processing and triaging but we don't use it for detection. We potentially could, but it's not what the product is meant for.
How has it helped my organization?
The visual playbook editor updates that they released have been absolutely instrumental because the old editor was impossible to look at for most of the time. It made my eyes bleed. I still have to look at it from time to time.
Splunk does provide substantial value.
It definitely does reduce our mean time to resolution through the enrichment details that it provides. Inputting your facts and details of the things you do not want to see with the events coming into it and easily filtering down off of that is one of the main value drivers outside of phish removal.
What is most valuable?
The most valuable feature is the API connector, depending on how it's formatted and who made the actual app offering for it. The REST API is my favorite component. It's very easy to use. The filters are also really valuable. Those are the two primary features but I enjoy using the rest of it.
What needs improvement?
SOAR is probably the most unreliable product Splunk has and that's because most of it is content driven from what you put into it. There are certain parts of it that have a little bit of difficulty at volume too. It's always changing. There is new stuff coming out for it that's going to make it a little bit better, but it does have some drawbacks.
It's specifically geared for SOC and not broader automation. The artifact filtering that's forced on everything inside the platform is pretty awful. It's for a subset of active playbooks which, out of the two hundred that we own, I think three or four of them are active, but we have to play with that setting for each one of them.
Every block should also have that option specifically because if you're not doing the artifact filtering on the front end, it's not good.
We've had lots of processes that have been victim to filtering not working appropriately at scale. It's hard to actually track down and trace because we can't reproduce the issues that we see in our testing environment or in production. That was two minor versions ago. It might have changed, it might not have, but we don't have a lot of trust in that feature.
UI elements like interacting with our analysts are near impossible. Finding stuff on the actual dashboard is really impossible most of the time. One example is that the timeline takes up three-quarters of the screen, but not a single person uses it because you have to individually set the container, the artifacts, and the actions to a specific attribute field that's really difficult to correlate to the actual events you put into it. The artifacts are really weird too because they're not traditional forensic artifacts. You shouldn't be able to change the value of an actual artifact. It was in that capacity but we also use it for that purpose in the platform.
For how long have I used the solution?
My company was one of Splunk's first five customers. I have been using it for the last three years.
What do I think about the stability of the solution?
I've only crashed SOAR a few times and it was my fault. If you have a production environment that's been running for a month or two and you have a few thousand events in it, if you mess up your query when you're trying to ask it a question and you do page size zero, it will just give you things on it, and it will crash it. That's a fun thing, but you shouldn't do that in general. That was a mistake on my part. Generally, it is very stable and available as most of the issues are usually the fault of the vendors that it's talking to, but that's with any platform.
What do I think about the scalability of the solution?
Scalability is interesting. Some of the assets do choke each other out. There is a cyclical lock thing that we had to fix on our inside. We have a CrowdStrike app, and we give it a file and ask it to do something and it goes great. It tells us that the default wait time is fifteen minutes, and there's only one of me. But there are five processes competing for that, and you get a giant backlog. We had to make our own custom app to get it later.
We have about fifty users on SOAR and a few hundred playbooks. Our environment is fairly large in terms of standard customers.
How was the initial setup?
I didn't do the initial integration, it was many years ago but we do deployments with the platforms team because we have the experience.
We have it down to a pretty good science right now because platform science does a really good job of automating the steps that go into setting up the server and whatnot. One good thing about the SOAR connectors that we have in the apps is the ability to save states and for apps just to self-heal. That has been really helpful because things go down from time to time and we don't have to worry about it because there's a second or third process that's going to pick it up.
What's my experience with pricing, setup cost, and licensing?
I have heard they are changing pricing, not possibly for the better. In comparison to the other vendors we looked at, they're all in the same ballpark of what they should be billing on. SOAR makes the most sense out of all of them, in terms of the billing factors.
Which other solutions did I evaluate?
We are looking at other platforms currently to compare areas. Splunk's editors are exceptionally better to look at. Visually, it's easier to find things and configure them.
There is more capability out-of-the-box for doing typical data transformation that you don't have to write too much code for, which is really nice. The code blocks have annotations in them. So when you actually open and look at what you worked on, four or five months later, you have your notes right there in the same place where it runs, which is really handy.
It's also just built for broader automation and it's all more HTTP, actions-based. Instead of having to build a connector, then put that on GitHub and install that in your platform, you can define an endpoint with credentials and you can do the same thing with SOAR. It's encouraged to do it with the actions and assets, which can be beneficial depending on what the product is.
If we do continue using SOAR, I think we're going to default to using more HTTP actions and stop using too many assets because it's a bit of a burden to create one, especially if out-of-the-box the actual configuration doesn't do what we need it to.
One example of this that we have is the request tracker app that we use for all of our tickets. When you ask it for the ticket information, it will return the metadata on it, nothing inside the actual ticket. That's a fork we have to create. It didn't actually do the basic product functionality that the vendor should be providing.
We also find that the vendors don't always keep the SOAR connectors updated. Sometimes they'll update the associated API, and then their connector will stop working because they're on different versions, and then we have to force our own fix on that. They usually make a SOAR connector just to say that they have one, but they won't put too much effort or thought into it.
What other advice do I have?
I'd probably rate the functionality an eight or a nine out of ten. I would give the UI a four out of ten. I would rate general Splunk SOAR a seven out of ten.
Disclosure: I am a real user, and this review is based on my own experience and opinions.