Vertica Review

I like the clustering aspect with the share-nothing mentality. I also value the ease of maintenance.


Valuable Features

The biggest, most valuable feature for us is the clustering aspect with a share-nothing mentality. Most clusters usually require their own shared storage, shared subnet, etc. and this becomes a pain and a nightmare to maintain.

The second most valuable feature is that it's very easy to maintain. It's a breeze once you know how to handle it with your scenario in mind.

Improvements to My Organization

Loading raw data and leveraging column store to quickly aggregate the values as well as run a general analysis were the biggest improvements we found. Before, we had to scrub the data or reformat, load it, possibly scrub it some more, and then run the first set of analysis, and so on.

With Vertica, we were able to combine some of these steps, such as loading gzip data directly into the table and leveraging R in Vertica to run all of the analysis.

Room for Improvement

Developer Tools - Vertica really needs some kind of IDE plugin for a system such as Eclipse or IntelliJ. Developing external functions in Vertica can kind of be like shooting in the dark sometimes. Also, an improved monitor or monitoring with alerting built-in that actually works would be a welcome addition.

They truly need a Python or some script that can handle all of the low-level system changes for you and find out how the customer has heavily modified their nodes before the install. Some automation here would help a lot.

The product overall is a great product, however management tools as well as monitoring tools are lacking. The product does, however, offer a lot of information in the form of system views and tables, but most of the data is hard to translate with out the help of their support team.

Use of Solution

I have used HP Vertica in multiple companies over the last four years. We currently have it running on a three-node Centos cluster and a six-node Centos cluster.

Deployment Issues

There have been no issues with the deployment.

Stability Issues

There have been no issues with the stability.

Scalability Issues

We have had no issues scaling it for our needs.

Customer Service and Technical Support

Like everything else HP has support for, the support is very poor. You normally have to threaten to leave, not buy support renewals, or call your sales rep to talk
to anyone who knows anything about the product. The community normally knows more than support and most of my questions or issues were resolved by searching the old community boards while I wait for overseas support to ask me to send them the logs again for the 50th time.

Previous Solutions

I have previously tried SQL PDW, Mongo, Cassandra for alternatives. Even though all of those products are in different landscapes, the Vertica column store ended up being the best thing that was needed.

Initial Setup

It is straightforward if you read the documents and have mid to senior-level knowledge of Linux. Non-Linux admins will find the setup complex and cumbersome since most are Windows admin and they want point-and-click.

Implementation Team

We implemented through our in-house team. You need to read the docs, then read them again, and then make yourself a cheat sheet. Once you have done the setup for a two-node cluster, do some Research and Development before taking the time to do a large production cluster or buy the license.

ROI

ROI is great compared to the previous solution, SQL Server.

Pricing, Setup Cost and Licensing

TCO is much lower given the Linux OS and the fact that Vertica is licensed by data size and not node count. The best advice for licensing is to make sure you have a proper data retention policy in place and well-documented as well as some growth expectations before buying. Following this, it will make sure you don't over or under buy.

Other Advice

If you are not Linux savvy, find a person that is. Make a cheat sheet with the commands and/or steps for your environment. If you are in the cloud, make sure to understand the networking aspect is completely different in AWS from it will be in your local data center. Failure to plan is planning to fail with Vertica implementation, and try not to mess up the spread as it's a pain to fix. If you read the documents, you will see what I am talking about.

Disclosure: I am a real user, and this review is based on my own experience and opinions.
Add a Comment
Guest
Sign Up with Email