January 2018 Summaries

In this GreyBeards On Storage podcast episode, the hosts discuss high performance databases and their required storage with Brian Bulkowski, Founder and CTO of Aerospike. Topics covered include database performance tied to storage (or memory), write IO performance, Intel Optane memory, and data consistency in databases. The full podcast runs for approximately 44 minutes and can be accessed on the Silverton Consulting site along with a detailed writeup and download option.

The text discusses the potential limitations of using Amazon Web Service's (AWS) DynamoDB, a popular choice for many enterprises. It suggests that while DynamoDB is a solid product, it may not be suitable for all business needs and could lead to cost overruns and technology limitations. The author recommends considering Aerospike as an alternative database solution, especially if companies are experiencing issues with DynamoDB. They provide a whitepaper titled "5 Signs You Should Consider Aerospike Over Amazon DynamoDB" for more information on this topic.

Companies often select databases based on initial impressions and convenience, but may later face cost overruns and technical limitations that hinder productivity and put the business at risk. This is a common scenario with Amazon Web Service's (AWS) DynamoDB, which may not be the best fit for all uses due to its limitations. As an unbiased assessment, Aerospike has conducted costing and benchmark-based testing to provide insights into when businesses may have outgrown DynamoDB, highlighting potential areas of concern that can help inform informed decisions about database selection.

The Meltdown and Spectre CPU bugs require OS and cloud hypervisor patches to ensure data security, but these patches can cause performance issues such as increased CPU use and higher latency. Aerospike has seen few impacts on its customer base due to its low CPU and system call usage. This blog provides guidance for monitoring, analyzing, and mitigating the performance aspects of the required security patches in various environments, including Amazon cloud and bare metal/hybrid cloud deployments. Recommended actions include monitoring latency and CPU load, upgrading kernels, optimizing networks, enabling hugepages, and considering disabling PTI patches for non-personally identifiable data.