Validate application and infrastructure performance under load
Identify code issues or system bottlenecks early on
Compare different architectural options on performance and cost dimensions
Create sizing and guidelines for applications
Helped a leading cloud NAS solution provider develop correct sizing guidelines for their product in AWS and Azure...Read More
TPC HCI is a standards based test for Hyper-Converged Solutions. It is a very complex test and involves setting up multiple application tiles...Read More
Customer is a leading hyper-converged infrastructure product vendor offering solutions for varied workloads such as VDI...Read More
Helped a leading cloud NAS solution provider develop correct sizing guidelines for their product in AWS and Azure.
Users reported performance issues with the product. These issues arose mainly due to users choosing wrong instance types or volume types in the absence of clear guidelines.
Based on understanding of product functionality. Broad matrix was defined - multiple instance types in AWS & Azure, volume types (e.g. Magnetic, GP SSD, Provisioned IOPS, S3), different types of workloads (block sizes, read/write mix, random/ sequential).
FIO based load generation, Monitoring Scripts, Data Management & Analysis using Elasticsearch.
Tests uncovered multiple performance issues arising from both inherent product issues and also due to limitations imposed by instances and volumes. As a result of the sizing, MapleLabs was able to put together a sizing guide that recommended the right choice of instances & volume configuration for desired performance.
TPC HCI is a standards based test for Hyper-Converged Solutions. It is a very complex test and involves setting up multiple application tiles (each tile consists of 1 Application VM and 2 DB VMs) on the cluster, scaling up load until the cluster performance saturates.
Benchmark multiple configurations of HCI with TPC-HCI workloads and more importantly, understand bottlenecks, tune the system and extract peak performance.
100s of test were run, analyzed and compared with each other. Data was sliced and diced to uncover causality and insights. The analysis was made possible by an excellent test management system built using MapleLabs’ SnappyFlow product that served as both a test repository and analytics system.
HCI is a complex system with a large number of control variables. In order to correctly and efficiently test the system, it is important to change the variables in the right sequence. MapleLabs created a very good test strategy as a result of deep understanding of the domain and experience in performance test.
MapleLabs was able to tune the system to reach its peak performance with minimal iterations. While doing so,several system and component issues were uncovered that were important inputs to the product designers.
Customer is a leading hyper-converged infrastructure product vendor offering solutions for varied workloads such as VDI, enterprise applications, database applications,
file services etc.
Create a reference design for Oracle single node and Oracle RAC for multiple HCI models that highlighted best practice configurations and load vs performance characteristics.
MapleLabs was involved in conducting performance benchmark tests to ensure customer’s hyper-converged products are suited to run complex Oracle applications. The project involved developing optimal hardware configuration / sizing and validating the same by executing performance tests. MapleLabs used Oracle Swingbench and Oracle SLOB workload test suite to run scale-up and scale-out using different VM sizes and multiple VMs running simultaneously. In addition, we tested database backup, restore and recovery features to measure throughput and elapsed times.
Solution enabled customers to rapidly deploy Oracle application by eliminating engineering and validation processes that are usually associated with deployment of enterprise solutions.