It is smoking fast. Analytical queries are usually run on multiple rows of information within the same column, instead of OLTP queries, which tend to process data row-by-row. Performing a high number of updates leads to performance degradation with time, until a VACUUM operation is manually triggered just like PG. To remain optimal, you want to continually revisit your cluster setup and tweak the quantity and sort of nodes. Redshift does not provide any customized function or array support. It is among the most cost-effective solutions readily available and allows for integration with many favorite BI tools.



Amazon Redshift costs under a tenth of the majority of traditional data warehousing solutions and is optimized for datasets that range from a couple of hundred gigabytes to petabytes. You can set up a small system and try it, and then resize larger for production.

Just like using Postgresql

Redshift utilizes precisely the same syntax as PostgreSQL. It can search through vast amounts of data and provide complex analytic results in shockingly short query times. Redshift offers you a choice to use Dense Compute nodes that are SSD based.


Redshift offers you a choice to use Dense Compute nodes that are SSD based and all handled by the provider. As with the majority of AWS services, you cover the services which you apply.

Ability to rebalance

In the event you do want to balance the data, there’s a rebalancing utility available. You’re able to collaborate employing precisely the same data across multiple Redshift clusters.


Database Redshift Performance Tuning Incident


Being a database administrator (DBA) is definitely not an easy job. It involves poring over data that may seem incoherent, constant confrontation with developers and frequent calls from customer service department. A DBA’s job description usually involves setting up servers, managing servers especially during initial server optimization and custom tweaking of the Redshift server to provide the best customer service possible.


Things seemed to be going on fine when he noticed that his phone had begun to ring. He realized that he hadn’t really fixed a specific time with her but wondered if it was not too early in the day for the visit, but then who was he to complain, he would only just need to speed up the cleaning process. A look at the screen of his phone, however, showed that it was his I.T manager and not Karen that was calling. With the I.T manager calling him on a day like this, it could only mean one thing and as he had suspected, it meant trouble.


Customers were calling to complain about the server and something had to be done about it asap. You could sense from the transfer of aggression in his tone that he was not having it easy with the Customer Relationship’s manager who in turn was not having it easy with the customers. Brad knew better not to suggest remote troubleshooting, pissing off his I.T manager on a day there was a crisis did not seem like a good idea and if the scenarios described is true, he knew that he had no choice but to go because he was the most capable hand to solve a problem of this type.

Seated on his seat, facing the large screen in front of him, he settled down to work.


One glance at the “cfg file” on the monitor showed him there was clearly a high server load, what was not easy was how to identify the cause. Just then the I.T manager came in accompanied by the customer relationship manager. “Brad, we didn’t know you had arrived”, queried the I.T manager. “I just go in”, he responded. After listening to what they had to say, what they had done to try to sort out the problem, he realized that the whole hierarchy of the company was looking up to him once again to save the day. Good customer experience was one of the policies of the company and everything had to be done to get the system backup and fast too. Every second counted and meant money.



From experience, problems like these were difficult to pinpoint even after a proper performance recording and analysis to identify slow worksheets, queries, and long render-duration. His first action was to look out for fragmented tables since this was a common reason for load spikes so as to optimized the ones that were fragmented, he also had a look at the read/write rate even though the server had more than enough HDDs arranged in a RAID array with good speed expected.


He then checked for bottlenecks in the memory, disk I/O, network and CPU resource. Nothing seemed out of normal. More checks showed that the %WA (Wait Average) was not the one utilizing the highest CPU%, which suggested that it was not an I/O induced bottleneck. He was, however, able to narrow the problem down to the growth in database size, and fragmentation of the tables as that occurred.


At the end of all the analysis checks, he was convinced he needed to do a performance tuning on the system. This can sometimes involve configuration changes to hardware, software and network components depending on what the problem was. He decided to use the opportunity to test one of the tools they recently acquired in carrying out the performance tuning.


Luckily for him, he did not need to do indexing as that had been done during their last maintenance. He then conducted query optimization by reducing the number of joins, SQL client code optimization, enabled InnoDB to handle high numbers of concurrent connections and blocked a suspicious process by an unidentified user who had an abusive connection to the system.


After several hours of coding, tuning, fine-tuning and testing, he noticed a significant improvement in the response time of the systems. The I.T manager and the customer relationship’s manager were glad to have the server back up. He decided to linger around for a while to further test the system and use the opportunity to document the log of the activities he had carried out.


With the work at the office done and the date with Karen confirmed, he decided he was going to leave unannounced to avoid getting a new assignment.


When she finally got in, he told himself that if he was going to impress this lady, he needed more than just Redshift performance tuning skills.



Help Managing AWS Cloud SRE for hire remote part-time