When you read my earlier blog, “Netezza is making a comeback”, from this blog series, did you think to yourself, “Where did it go? I never left.” If you have ever had a chance to use Netezza in your data estate for warehousing, high-performance INZA push down capabilities, or the simplicity of ownership, you know how lucky we have it. Netezza covers a lot of sin was a mantra – where Netezza was able to cover up poor data structures and queries with forgiving data access.
This continues today with the new generation of Netezza, Netezza Performance server, available on-prem, in any cloud, and in hybrid postures. It includes the same forgiveness, simplicity, and power, overcoming some less than optimal queries and access.
Netezza “administrators”(I am using quotes to indicate a lot of the admin function is not required with Netezza) enjoyed the benefits of not needing to prepare the system for data warehouse workloads at installation and configuration. Netezza came preconfigured to load data fast and read data faster.
The administrators did not have to worry about extracting the most performance out of a generic install by adding partitions, complex indexing, advanced data modelling, or query run time analysis. All of that work was performed by the system. Now, for advanced installations (which account for 10% of our clients), we go ahead and configured the product to get even more out of the system; however, this is an exception, not a rule.
There are a few more things we will need to concern ourselves with when we move into this new containerized world, and the OS and database are able to work anywhere. There are a couple of additional processes that we must ensure are running, but the core (data loading and data reading) remains the same.
Buy or Build?
Compared to most relational databases, Netezza and Netezza Performance Server is specified and tuned for analytics workloads. Most RDBMS installations start with a generalized and average basis for installation and configuration. Setting up partitions, disk, location or data, logs, indexes, striping, and database segmentation all must be designed. All of this design must be documented along with all the changes made to the configuration and maintained for the next system set up and to support analysis for long running data load or data reads.
Most RDBMS are naturally designed for online transactional processing and not analytics. There are many things that must be overcome to install, configure, and tune them into what Netezza provides out of the box.
The cloud options that are being offered are a good and fast way to get analytics databases set up and established quickly. Since most of them are derivations of Postgress, the conversion and changes to data structures is minimal. However, the changes to the system and how data integration and reporting tools interact with the analytic database do require consideration in a new hybrid environment and in cloud-first topographies.
Customers moving from the traditional on-prem data source (ERP, HRIS, CRM etc.), on-prem data integration, on-prem analytics database, on-prem reporting, and on-prem consumption need to review the solution and ensure the code is prepared for the changes made by moving one or all of the above to the cloud. If all the data being manipulated rests on-prem or all in the cloud, the solution becomes easier to set up. By having hybrid – for either the planned end state or during the transition of one or more on-prem environments to the cloud – will need a lot of review and rationalization.
We have examples of clients who jumped into this new cloud paradigm without doing such analysis. Data loads took longer, data reads took longer, and some reports did not even render. Additionaly, the compounded costs of simply moving the same code into this hybrid solution and some learning of what to shut off when and where resulted in 10X expected costs monthly when combining the database subscription with the cloud subscription (for most of these cloud options, you will have two bills/credits to manage – one from the database vendor and one from the platform as a service cloud vendor).
And what if you want to take back the solution and the cloud was not worth the hype, proposed cost savings were not realized, and/or the performance was not up to standards set by the original system? There is not a way back for those vendors.
This is why we are excited about the Netezza Performance Server being on Cloud Pak for Data supported by RedHat OpenShift. The container abstract allows the OS and DBMS to run the same, regardless of where the container is installed. No refactoring, no rework, and no effort.
Now, this doesn’t mean that when moving workloads to the cloud you should not consider refactoring – it just means it is not required before the value of the cloud can be realized. Hybrid solves a lot of issues we have had with a on-prem or cloud-only solutions.
Once a year, quarter, or month – large batch or exceptional updates or reads can be performed on NPS in the cloud, and once the job is completed the system can scale down.
Data science projects need all the power and data they can get, but they can take a long time, running over a number of days. To avoid impacts to production, the workload can be performed on NPS in the cloud.
Data replication, back up, and disaster recovery. The system is the same for each, and they are all containerized – the data can be replicated, scripted to back up and restore, and, if that back up is available, it can be restored to any NPS system.
Hopefully you can see how easy we have it – Owning and using the Netezza appliance and/or the Netezza Performance Server is a great solution for business that are not dedicated to configuring and constantly adjusting their analytics data stores. It’s excellent for a company that wants these initiatives to be fast, easy, and cost effective.
I am a technologist at heart, and I love to hear stories about building high performance compute, GPU, accelerators, and complex cloud-based solutions that have unmatched performance (often at a high cost). As a business leader, I am more and more drawn to solutions that just work. The ones that a company can deploy and use without investing in the care and feeding and re-architecting of the solution.
If you have had a chance to work with Netezza in the past, I encourage you to explore how Netezza Performance Server could compliment your data estate. If you have not had a chance to work with Netezza in the past, I encourage you to get a cloud instance set up so you can experience the speed and simplicity the rest of us all enjoy!