Cross rack network latency in AWS

I have VMs running on bare-metal instances. Each bare-metal instance is in a separate rack by design (for fault tolerance). The bandwidth is 25GbE however, the response time between the hosts is so high that I need multiple streams to consume that bandwidth.

Compared to my local on-prem lab I need many more streams to get the observed throughput close to the theoretical bandwidth of 25GbE

# iperf StreamsAWS ThroughputOn-Prem Throughput
14.8 Gbit21.4 Gbit
29 Gbit22 Gbit
418 Gbit22.5
823 Gbit23 Gbit
Difference in throughput for a 25GbE network on-premises Vs AWS cloud (inter-rack)

How to run vdbench benchmark on any HCI with X-Ray

Many storage performance testers are familiar with vdbench, and wish to use it to test Hyper-Converged (HCI) performance. To accurately performance test HCI you need to deploy workloads on all HCI nodes. However, deploying multiple VMs and coordinating vdbench can be tricky, so with X-ray we provide an easy way to run vdbench at scale. Here’s how to do it.

Continue reading

Why does my SSD not issue 1MB IO’s?

First things First

https://commons.wikimedia.org/wiki/File:CDC9762-smd-drive.jpg
CDC 9762 SMD disk drive from 1974

Why do we tend to use 1MB IO sizes for throughput benchmarking?

To achieve the maximum throughput on a storage device, we will usually use a large IO size to maximize the amount of data is transferred per IO request. The idea is to make the ratio of data-transfers to IO requests as large as possible to reduce the CPU overhead of the actual IO request so we can get as close to the device bandwidth as possible. To take advantage of and pre-fetching, and to reduce the need for head movement in rotational devices, a sequential pattern is used.

For historical reasons, many storage testers will use a 1MB IO size for sequential testing. A typical fio command line might look like something this.

fio --name=read --bs=1m --direct=1 --filename=/dev/sda
Continue reading

How to identify SSD types and measure performance.

Thomas Springer / CC0
Generic SSD Internal Layout

The real-world achievable SSD performance will vary depending on factors like IO size, queue depth and even CPU clock speed. It’s useful to know what the SSD is capable of delivering in the actual environment in which it’s used. I always start by looking at the performance claimed by the manufacturer. I use these figures to bound what is achievable. In other words, treat the manufacturer specs as “this device will go no faster than…”.


Identify SSD

Start by identifying the exact SSD type by using lsscsi. Note that the disks we are going to test are connected by ATA transport type, therefore the maximum queue depth that each device will support is 32.

# lsscsi 
[1:0:0:0] cd/dvd QEMU QEMU DVD-ROM 2.5+ /dev/sr0
[2:0:0:0] disk ATA SAMSUNG MZ7LM1T9 404Q /dev/sda
[2:0:1:0] disk ATA SAMSUNG MZ7LM1T9 404Q /dev/sdb
[2:0:2:0] disk ATA SAMSUNG MZ7LM1T9 404Q /dev/sdc
[2:0:3:0] disk ATA SAMSUNG MZ7LM1T9 404Q /dev/

The marketing name for these Samsung SSD’s is “SSD 850 EVO 2.5″ SATA III 1TB

Identify device specs

The spec sheet for this ssd claims the following performance characteristics.

Workload (Max)SpecMeasured
Sequential Read (QD=8)540 MB/s534
Sequential Write (QD=8)520 MB/s515
Read IOPS 4KB (QD=32)98,00080,00
Write IOPS 4KB (QD=32)90,00067,000
Continue reading

Quick & Dirty Prometheus on OS-X

How to install Prometheus on OS-X

Install prometheus

  • Download the compiled prometheus binaries from prometheus.io
  • Unzip the binary and cd into the directory.
  • Run the prometheus binary, from the command line, it will listen on port 9090
$ cd /Users/gary.little/Downloads/prometheus-2.16.0-rc.0.darwin-amd64
$ ./prometheus
  • From a local browser, point to localhost:9090
prometheus web-ui

Add a collector/scraper to monitor the OS

Prometheus itself does not do much apart from monitor itself, to do anything useful we have to add a scraper/exporter module. The easiest thing to do is add the scraper to monitor OS-X itself. As in Linux the OS exporter is simply called “node exporter”.

Start by downloading the pre-compiled darwin node exporter from prometheus.io

  • Unzip the tar.gz
  • cd into the directory
  • run the node exporter
$ cd /Users/gary.little/Downloads/node_exporter-0.18.1.darwin-amd64
$ ./node_exporter
 INFO[0000] Starting node_exporter (version=0.18.1, branch=HEAD, revision=3db77732e925c08f675d7404a8c46466b2ece83e)  source="node_exporter.go:156"
 INFO[0000] Build context (go=go1.11.10, user=root@4a30727bb68c, date=20190604-16:47:36)  source="node_exporter.go:157"
 INFO[0000] Enabled collectors:                           source="node_exporter.go:97"
 INFO[0000]  - boottime                                   source="node_exporter.go:104"
 INFO[0000]  - cpu                                        source="node_exporter.go:104"
 INFO[0000]  - diskstats                                  source="node_exporter.go:104"
 INFO[0000]  - filesystem                                 source="node_exporter.go:104"
 INFO[0000]  - loadavg                                    source="node_exporter.go:104"
 INFO[0000]  - meminfo                                    source="node_exporter.go:104"
 INFO[0000]  - netdev                                     source="node_exporter.go:104"
 INFO[0000]  - textfile                                   source="node_exporter.go:104"
 INFO[0000]  - time                                       source="node_exporter.go:104"
 INFO[0000] Listening on :9100                            source="node_exporter.go:170""
Continue reading

SQL Server uses only one NUMA Node with HammerDB

Some versions of HammerDB (e.g. 3.2) may induce imbalanced NUMA utilization with SQL Server.

This can easily be observed with Resource monitor. When NUMA imbalance occurs one of the NUMA nodes will show much larger utilization than the other. E.g.

Imbalanced NUMA usage by SQL Server.

The cause and fix is well documented on this blog. In short HammerDB issues a short lived connection, for every persistent connection. This causes the SQL Server Round-robin allocation to send all the persistent worker threads to a single NUMA Node! To resolve this issue, simply comment out line #212 in the driver script.

Comment out this line to work-around the HammerDB NUMA imbalance problem.

If successful you will immediately see that the NUMA nodes are more balanced. Whether this results in more/better performance will depend on exactly where the bottleneck is.

Balanced NUMA usage by SQL Server