Skip to content

Production Release of Data At Scale Hub

Compare
Choose a tag to compare
@jamesbak jamesbak released this 30 Dec 20:33
· 5 commits to master since this release

We're excited to announce that this latest release is our v1.0 - DASH is now a production release! With this release, we believe DASH is appropriate for use in production scenarios.

The main new feature added in this release is a new Management API and Portal. Additional small improvements and bugfixes are also included.

Management API and Portal

DASH now exposes a new HTTP(S) WebAPI endpoint to remotely manage configuration of the deployment. Additionally, a new single page application can also be served from this same endpoint which can act as a management portal.

Current functionality exposed by the management api is limited to:

  1. Updating configuration - all of the configuration entries may be viewed and updated
  2. Provisioning of new scale-out storage accounts - if a new storage account is added to the configuration it will be automatically provisioned in Azure with the appropriate location and replication characteristics
  3. Software update - as new releases of the DASH software are published, they may be viewed in the Management Portal. Additionally, different versions of the software may be deployed without any downtime to the service

The api and portal are secured by Azure Active Directory. See this link for documentation on how to setup authentication: https://github.com/MicrosoftDX/Dash/wiki/Configure-Authentication-for-Management-API.

The Management Portal documentation is located here: https://github.com/MicrosoftDX/Dash/wiki/Dash-Management-Portal. The Management API is documented here: https://github.com/MicrosoftDX/Dash/wiki/Dash-Management-API

In the future, the Management API & Portal will be extended to provide monitoring of the Dash service itself as well as metrics for the overall performance of the virtual account.

Other Enhancements

  1. Support has been added to fully enable Azure Batch (https://azure.microsoft.com/en-us/services/batch/) to leverage the read replica feature of Dash thus improving scale for read throughput. Previously, Batch jobs had to perform their own I/O to gain the benefits of Dash. With this change, the native Azure Batch I/O library will be fully supported by Dash.
  2. Added support for Azure Storage versions 2015-02-21 and 2015-04-05. These versions clarified a previously ambiguous requirement for content-length in the string to sign. The new version is now unambiguous and is authenticated correctly by Dash.

Bug Fixes

  1. Fixed resource leak when forwarding/proxying requests to data account. Under extreme cases, the resource leak was observed to be sufficient to exhaust TCP/IP ports on the host server.
  2. Fixed multiple bugs associated with incorrectly encoding/decoding URIs containing special characters. In the worst case, the namespace blob exhibited a different name encoding that the associated data blob and could therefore not be resolved.

Let Us Know What You Think

Please feel free to drop the team a note on Github if you have thoughts about Dash or scalable storage generally. We're always looking out for new scenarios that improve the scalability and therefore applicability of Azure Storage.

Enjoy!