This walkthrough is an overview of the See It at Scale cluster. To access the cluster, you need to log in to the Qumulo web site and get your credentials. (The Log In link is at the upper-left hand of the screen.)
The See It at Scale cluster is a live QF2 cluster that consists of six QC104 nodes with 485.5TB usable (624TB raw) capacity. Over 3.2 billion files exist on the file system. Four Linux clients generate varying amount of IOPS and throughput, depending on the workflow.
Currently, there are three different workflows running on the cluster. 1. Continuous playback of over 6000 sequential 50MB .dpx image files to simulate a movie sequence. 2. Genomic processing of actual human genomic data. Raw .fasta files are processed and aligned in parallel. 3. Spark-based analytics where the four clients use Spark to count the words in a 125GB text file.
The workflows automatically cycle. You can identify which workflow is running by examining the file structure. No workflow takes longer than 10 minutes, so if you’re patient you’ll be able to see them all.
To get started, log in with the credentials you received. You will see the dashboard. Here is an example.
The dashboard gives an overview of cluster activity over the last 24 hours. IOPS and throughput are graphed at the bottom, with reads in purple and writes in green.
Varying amounts of cluster activity are shown. The detailed zoom window can be adjusted by grabbing the blue slider and dragging either arrow to expand or contract the detail. It can also be scrolled over different time periods from the last 24 hours.
In this section, you’ll look at some of the analytics that are available. If you want more help, you can look at the tutorials and videos on the Try QF2 Now page.
Point to the Analytics tab and click Integrated Analytics. Here is an example of what you’ll see.
The screen displays, in real time, the most “interesting” things happening with the cluster. Throughput and IOPS (both file and metadata) are represented as bars in the directories where activity is occurring. If you click in one of the directories, you’ll see that the most active clients and directories are displayed on the left.
You can focus on specific types of access by unchecking the boxes on the upper right. Try and change the display so only Metadata activity is represented.
Note how quickly all this information becomes available. QF2 doesn’t need to do a treewalk to generate this information. Experiment and try clicking on other directories to learn more about the file system.
The capacity explorer lets you see which directories contain the most data.
Point to Analytics and click Capacity Explorer. Here is an example of what you’ll see.
Capacity trends gives you an historical view of how storage is being used over time.
Point to Analytics and click Capacity Trends. Here is an example of what you’ll see.
The selector lets you choose from trends over the last 72 hours, 30 days, or 52 weeks.
You can easily identify throughput hot spots.
Point to Analytics and click Throughput Hot Spots. Here is an example of what you’ll see.
Which directory has the most throughput? Is the throughput reads (purple)or writes (green)?
You can easily locate IOPS hot spots.
Point to Analytics and click IOPS Hot Spots. Here is an example of what you’ll see.
Note that both file and metadata IOPS are shown.
You can easily identify which clients are the most active.
Point to Analytics and click Client. Here is an example of what you’ll see.
The clients with the most activity are displayed. Click on any client and it will expand to show the most active directories.
Which client is most active and what directories is it accessing?
You can easily identify which paths are most active.
Point to Analytics and click Path. Here is an example of what you’ll see.
The most active paths are displayed. Click on a path and you’ll see the client that’s accessing it.
Feel free to explore the cluster and see how it handles the different workloads. Get a sense of how quickly you can get information. Can you get comparable information in real time with the system you’re using now? Don’t forget the tutorials page, which will show you how to set up a demo cluster of your own, either on your computer or in AWS.
We are always looking for new challenges in enterprise storage. Drop us a line and we will be in touch.
Enter a search term below