|
Article: Is Joyent Really 14X Faster than EC2 and Azure the "Fastest Cloud"? Questions to Ask About Benchmark Studies |
|
November 17 2011
|
Benchmark-based studies and headlines can be misleading. In this blog post we propose a few questions one might ask when reviewing such studies. We then present 2 case studies based on those questions.
|
|
|
Article: Encoding Performance: Comparing Zencoder, Encoding.com, Sorenson & Panda |
|
October 17 2011
|
This blog post provides a summary of analysis we conducted comparing transcode performance across 4 different cloud based encoding services.
|
|
|
Article: Do SLAs really matter? A 1 year case study of 38 cloud services |
|
January 15 2011
|
A year ago we signed up for services with dozens of different IaaS and PaaS cloud providers. We then setup monitoring on each of those services. Scheduled maintenance periods were excluded and we also attempted to manually confirm and document extended outages. This article examines the availability we experienced during one full year with each of these services, and compares that metric to the provider SLAs to determine if there is a correlation between the two.
|
|
|
Article: Benchmarking of EC2's new Cluster Compute Instance Type |
|
September 08 2010
|
In July 2010 Amazon released a new EC2 instance type, the Cluster Compute Instance or cc.4xlarge. This new instance type is primarily targeted for HPC workloads. The major distinguishing features between the cc.4large and other EC2 instances types are Hardware-Assisted Virtualization and 10 Gbps non-blocking network. This article summarizes 100 different CPU, IO and application benchmarks we ran on the cc.4xlarge and how the results compare to other EC2 instance types.
|
|
|
Article: Cloud Server Benchmarking: Encoding & Encryption |
|
June 28 2010
|
It is often difficult to compare application performance between different IaaS services. Varied performance terminology, underlying hardware, and multi-tenancy factors add complexity to problem. In this post we summarize the results of benchmarking 150 cloud servers in 20 public clouds. In it, we focus on 7 benchmarks related to encoding and encryption application performance and attempt to provide a comparable performance metric for each cloud server benchmarked.
|
|
|
Article: Disk IO Benchmarking in the Cloud |
|
June 05 2010
|
Storage IO performance varies significantly between different IaaS services. Factors that affect performance include storage type (local or networked), disk type (SATA, SAS, SSD), configuration (Raid), and multi-tenancy (shared IO). In this post we summarize the results of benchmarking 150 cloud servers in 20 public clouds. Using a physical server with SAS Raid 1+0 disks as a baseline, and the results of 7 IO focused benchmarks, we attempt to provide a comparable storage performance metric for each cloud server benchmarked.
|
|
|
Article: What is an ECU? CPU Benchmarking in the Cloud |
|
May 29 2010
|
It is often difficult to compare IaaS services because the terminology used to describe compute resources is very different. For CPU resources, EC2 uses ECUs, vCloud use vCPUs, KVM uses MHz, etc. In this post we summarize the results of benchmarking 150 cloud servers in 20 public clouds. Then, using EC2's ECU as a baseline, and results of 19 CPU focused benchmarks, we attempt to provide a comparable CPU performance metric for each cloud server benchmarked.
|
|
Content
