In Linux, once a shared drive is mounted on a machine as a directory, the files within it can be accessed without concern about their actual location. Normally a system administrator sets the mount points and the users just access the data without necessarily know where data is actually coming from.
However, one result can be that users may be storing or accessing data in suboptimal locations. They could frequent access files on slower storage than necessary or even in different data centres.
One of our customers frequently uses Breeze to get around this otherwise very time-costly scenario.
Using the Breeze Directories View
One of the most popular strings to Breeze’s bow is that it provides a full list of file dependencies for a user’s application. In the case of the customer mentioned above, the engineer then uses the Directories View to only view directories as he isn’t interested in specific files.
The engineer then connects to a PC on the cluster as the user who ran the job, and looks up where those directories are connected to on the network, making a note of the network IP address or name.
The next step is to take the list of network connections and convert them to network host names. From here he can use the company’s database to see where in the world the user is based.
If the hostname is in a different physical location to where the job ran, they are defined as Wide Area Network (WAN) mounts rather than Local Area Network (LAN) mounts, and this is reported back to users. The aim is to carry out the compute in the same location as the data.
Using Breeze makes this search and locate job much faster for this customer.
Although there will always need to be some customer logic used to look up actual server locations, we are currently looking to add information into Breeze about what mountpoints are being used. Once this is done, the customer’s time will be even further saved.
Get in touch if you’d like to tell us about other features you’d like to see in Breeze.