Vectra AI Review

Easy to deploy and maintain, gives us ML, AI, and custom detection options for rule detection, and saves storage cost and time


What is our primary use case?

In terms of deployment, we have one brain and seven physical sensors. We're currently working on deploying a large number of virtual sensors, but those aren't done yet. We also have a SIEM and an EDR.

How has it helped my organization?

There are a large number of difficult-to-manage devices on a network. Traditional security vendors do a great job of making sure that workstations and servers are properly protected, secured, and observed, but they fall short when we're talking about odd peripherals, such as printers, scan guns, tablets, guest devices, and things like that. That's what Vectra helps us see. I can't tell the number of employee guest phones that just show up on the network, and they're infected because they're not managed by us and people do things with their phones. Now, we're able to actually see those devices hit our internal LAN instead of our guest networks, and we can properly move them over, whereas earlier, we were blind. Now, we have some reasonable assurance that our internal tablets, scan guns, and things like that are not performing abnormal network behavior. So, that's what we use Vectra for.

We've got a centralized data center with a large number of physical locations throughout the country. So, our network is very distributed. It's very much like a campus. Vectra is really good at reducing the complication of deploying an NDR solution, and that really helps us because we have over 175 stores that we need to capture traffic from, as well as a number of sales offices, regular employee offices, and distribution centers distributed across the country. So, Vectra makes it really easy. We just drop or ship it over there, and it is up and running real quick once it gets there. Shipping takes longer than configuration. So, basically, our network is a centralized data center infrastructure with a large number of stores, distribution centers, and offices geographically dispersed around the country.

It provides visibility into behaviors across the full lifecycle of an attack in our network beyond just the internet gateway. We tap client to server, server to server, and client and server to internet traffic, and it does a good job. It doesn't have an issue with internal traffic. In terms of the full lifecycle of the attack, Vectra is not designed to interface with or inspect the host. So, we're not seeing host activity obviously. That's what our EDR is doing. Vectra does an okay job. If we get a weird detection, we're also able to see a large number of other activities that happened just before and just after the attack and relate those to it.

Before we deployed Vectra, we were not monitoring network traffic. So, there was definitely a need and a gap, and Vectra has filled it. We have reliable network logs that are readable, and it does a good job of doing a default set of detections for us. We're very happy with the gap that it has filled.

It has overall reduced the time to respond to attacks, especially with the PCAP function on the detection, where when it gets a detection, it PCAPs the session. So, we're able to get a lot of context to alerts that we were unable to get before we deployed this because we weren't doing a full PCAP. Because Vectra only PCAPs the session when it triggers a detection, we didn't have to deploy hundreds of terabytes of storage across our network. So, we saved a lot of money there. There are $50,000 to $100,000 storage cost savings because it only captures the full packet capture for traffic that triggers detections. In terms of time, it has saved hundreds of hours. I can't even explain how happy we are with the amount of time it has saved us. Imagine the amount of time it would have taken us to deploy to 175 stores plus dozens of distribution centers and dozens of remote offices. Even if it was just one hour per location for deployment, that makes it hundreds of hours. Vectra, with being so easy to deploy and so easy to maintain and administer, has saved us hundreds of hours just on deployment and standing up the environment alone. I am not counting the maintenance and administration that come along with the solution.

What is most valuable?

It does a reliable job of parsing out the logs of all the network traffic so that we can ingest them into our SIEM and utilize them for threat hunting and case investigations. It is pretty robust and reliable. The administration time that we spend maintaining it or troubleshooting it is very low. So, the labor hour overhead is probably our largest benefit from it. We spend 99% of our time in Vectra investigating cases, responding to incidents, or hunting, and only around 1% of our time is spent patching, troubleshooting, or doing anything else. That's our largest benefit from Vectra.

We've got machine learning and AI detections, but we also have the traditional ability to create our own custom detections and rules that are important to us for compliance. When we were demoing other vendors, a large number of vendors let you make your own rules, but they don't provide their own rules and ML and AI rule engine, or they provide AI and ML, but they don't allow you to make your own rules. Vectra is very nice in that sense. We have detection rules that Vectra provides that are very common to the security industry, such as whenever there's a major event like the SolarWinds event. Those rules get built and deployed for us really quickly. We can manage our own, but then we also have the ML and the AI engine. We really like that. It is one of the few platforms that we've found to be supporting all three options.

What needs improvement?

They use a proprietary logging format that is probably 90% similar to Bro Logs. Their biggest area of improvement is finishing out the remaining 10%. That 10% might not be beneficial to their ML engine, but that's fine. The industry standard is Zeek Logs or Bro Logs, or Bro or Zeek, depending on how old you are. While they have 90% of those fields, they're still missing some fields. In very rare instances, some community rules do not have the fields that they need, and we had to modify community rules for our logs. So, their biggest area of improvement would be to just finish their matching of the Zeek standard.

They could provide distributed endpoint logging capability. We have a lot of remote workers nowadays in the day of the pandemic. If they're not connected to our VPN, then we're not capturing that traffic. So, the ability to do the traffic analysis for endpoints that are distributed would be cool. I have no idea how they would do that. I'm not aware of a single vendor that does that, but it would be cool if they could do that. To my knowledge, that's not really possible with the amount of compute power it would take on endpoints. It would be ridiculous. They'd have to really invent something new and novel that doesn't exist today in order to accomplish that. If they do, that would be great. Because I'm a customer already, I would use it. 

Cost-wise, they're not cheap. They were definitely the most expensive option. Their licensing model is antiquated. We have to pay for licensing based on four different things. They need to simplify their licensing down to just one thing.

For how long have I used the solution?

We have been using this solution for around 18 months.

What do I think about the stability of the solution?

I'm very happy with it. In the 18 months, I cannot recall any outage. We keep up on all the patching and maintenance, and there have been very few bugs. The SaaS product Recall has always been there when we use it. Our on-prem version has never broke. It seems very stable.

What do I think about the scalability of the solution?

It has got no problem with scaling. We picked Vectra because it was able to scale up to our size fairly easily without scaling up the deployment and administration overhead. So, it scales really well. It has no problem handling our volume of data.

How are customer service and technical support?

Their technical support is pretty good. They're very responsive. Nine out of 10 times, they understand my problem. They're not perfect, obviously, but at the end of the day, I got answers for the few issues for which I've had to use support. I can only think of one instance where it was painful, and that's why I say nine out of 10 instead of 10 out of 10. The guy just didn't understand what I was asking, and about seven emails later, it got triaged, and the next guy figured it out. Other than that, the first person I email in at support is able to answer my question in that initial response or just one extra email.

Which solution did I use previously and why did I switch?

We did not use any similar solution. 

How was the initial setup?

We have a couple of SaaS-based products. We use Cognito, Recall, and Stream. Recall is their SaaS-based product where all the logs go into their hosted elastic search instance, which allows us to search and create custom rules and everything like that, and then we pull data from that environment into our on-prem environment. In terms of the deployment of the brain, that's all on-prem. All the sensors are on-prem obviously, but we do use Recall.

In terms of the effort involved in deployment considering that some of the pieces we use are SaaS-based, it was literally just a toggle switch and an API client and key in the interface, and then it was working. We had to wait for accounting to approve it, and it added a little bit more time to our deployment because of paperwork, but technically, it was pretty simple. We told them we wanted this, and by the time that we got our paperwork done, everything at their end was stood up and ready to go for us.

It does take two to three weeks for the brain to baseline and establish its ML baseline. The moment it was done with the two-week to three-week machine learning period, it was good. So, it started providing value after three or four weeks after deployment.

What's my experience with pricing, setup cost, and licensing?

Their licensing model is antiquated. I'm not a fan of their licensing model. We have to pay for licensing based on four different things. You have to pay based on the number of unique IPs, the number of logs that we send through Recall and Stream, and the size of our environment. They need to simplify their licensing down to just one thing. It should be based on the amount of data, the number of devices, or something else, but there should be just one thing for everything. That's what they need to base their licensing on. 

Cost-wise, they're not cheap. They were definitely the most expensive option, but you get what you pay for. They're not the cheapest option. I know that their prices scared away a couple of people who have demoed it in the past. Once they got their quote, they were like, "Well, see you later. We can't do this." So, that is an area that they come up short against other people.

Which other solutions did I evaluate?

We did evaluate other options. We evaluated rolling Bro or Zeek on our own. We evaluated Security Onion. We also evaluated Corelight and almost picked them. We also investigated a couple of solutions that are significantly more involved than Vectra, just like full managed solutions, but we decided not to do that.

The main reason for choosing Vectra over all the other solutions was twofold. One was the deployment time and routine administration costs. Its deployment was very simple. The amount of time it would take to deploy and configure was very low. The time it would take to maintain the environment was significantly lower than the other solutions and on par with Corelight.

The second reason for picking it up is that it allowed us to create our own detection rules. They build rules for us when there are major events, as well as they have the ML and AI engine. This was the only solution that was easy and fast to deploy and maintain, and that was giving us all three options for rule detection. That's why we went with them. Some of the solutions provided all three options, but they were a pain to configure and maintain, and some of them were easy to deploy and maintain, but they didn't provide all three options.

What other advice do I have?

It is pretty straightforward. Plug it in and use aggregators in front of the sensors to aggregate multiple tap sources into a single sensor. The sensors can handle it. They de-duplicate everything. There is no need to purchase a sensor for every tap. Truncate all that traffic into an aggregator and have it come out one feed into the sensor. There is no issue there with the Vectra sensor being able to carve out all that. They're powerful enough to do that. Vectra recommends that. So, if someone is purchasing Vectra, they're going to hear that from them. With Vectra, you're picking reliable and fast among cheap, reliable, and fast.

In terms of Vectra's ability to reduce alerts by rolling up numerous alerts to create a single incident or campaign for investigation, we do not generate a lot of incidents. We're pretty quick off the gun on detections. We're responding to detections before subsequent detections are detected and become an incident. We maybe get one incident a week, so I don't know if I can comment on that effectively.

We don't use privileged account analytics from Vectra for detecting issues with privileged accounts. In terms of its detection model for providing security around things like Power Automate or other anomalies at a deeper level, we don't use Power Automate, but we use their anomaly detection, and it is very interesting. While it always does provide us something interesting to look at, more times than not, it is our IT admin who does anomaly detection. So, we learn a lot, and it brings odd things to our attention, but with anomaly detection, it has usually been our IT admin.

In terms of Vectra helping our network's cybersecurity and risk-reduction efforts in the future, I'm hoping that one day, we can achieve even client-to-client inspection. Vectra should stay up with the times, and they shouldn't start coasting, which I don't see at all. They fill a good gap, and they do that well. We're just going to leave them filling that gap until the time comes where that is no longer a need, which I don't foresee. So, I don't know if they're going to do anything more than inspect network traffic and provide us an alerting engine on anomalous or malicious network traffic. That's their niche, so that's what they're going to do, probably just more of it. As we grow, we'll deploy more Vectra sensors to capture that extra traffic. I see them scaling very well.

I would rate this solution a solid eight out of 10. It loses a star for not adhering to Bro Logs in my book, and there is no perfect 10.

**Disclosure: IT Central Station contacted the reviewer to collect the review and to validate authenticity. The reviewer was referred by the vendor, but the review is not subject to editing or approval by the vendor.
More Vectra AI reviews from users
...who work at a Financial Services Firm
...who compared it with Darktrace
Learn what your peers think about Vectra AI. Get advice and tips from experienced pros sharing their opinions. Updated: August 2021.
534,768 professionals have used our research since 2012.
Add a Comment
ITCS user
Guest