New cloud storage services for Microsoft Azure at Ignite 2018

At its Ignite 2018 conference, Microsoft announced Data Box hardware appliances that will allow cloud ingestion to tens or hundreds of TB, the availability of its Avere vFXT cloud NAS and cloud storage based on NVMe flash.

The first of these, Azure Data Box, allows migration of data in volume to the Microsoft cloud through a series of physical appliances. This is the Redmond giant’s response to Amazon Web Services’ (AWS) SnowBall Edge. It allows enterprises to dump large quantities of data into Azure without having to traverse the WAN.

The Data Box offering comprises two appliances deliverable by rapid courier service, dubbed Data Box and Data Box Heavy, with capacities of 80TB and 800TB useable respectively.

The appliances come ready to connect to the enterprise network by multiple Ethernet interfaces that range from 10Gbps to 40Gbps. The first comes in ruggedised server format while the second comes as a mini rack on wheels. The two appliances offer file access as well as via Microsoft’s Rest Azure Blob application programming interface (API).

Once Data Box hardware is connected to the network and configured, it is possible to transfer data in volume, including backups, archiving and unstructured data. When the transfers are completed the appliances are couriered back to Microsoft, which puts the content into Azure Blob or File.

Prices are €210 for the standard Data Box service and €1,686 for the Heavy version, with a credit of 10 and 20 days respectively for data transfer.

Every day on top of those 10 or 20 days used for filling the appliances is billed at €12.65 and €42.17 respectively. To this cost, delivery charges must be added (€95 for Data Box and €1,500 for Heavy).

The Data Box offer also includes two connected appliances – the Data Box Gateway and Edge. As the name suggests, the Gateway allows data transfer to Azure using protocols in use in the enterprise. It supports server message block (SMB) and network file system (NFS) as ingest protocols to Azure Blob Storage.

The Data Box Edge has similar functions to the Gateway device but is conceived of as a local cache for internet of things (IoT) applications. The appliance, which is in the form of a 1U Dell-produced server, has a small amount of local storage capacity to collect, for example, collected telemetry data and process it locally.

In this way, Microsoft aims to support Azure Machine Learning on the local device with specialised field-programmable gate array (FPGA) cards made available for Data Box Edge to support this.

The FPGA card is the outcome of developments in Microsoft’s Project Brainwave, whose mission is to allow real time machine learning functions. This is similar to Google’s tensor Processing Unit, which is also dedicated processing hardware for artificial intelligence (AI).

Microsoft has not explained if the card is the one based on Intel’s Stratix10 FPGA like the prototype it showed a year ago at the HotChips conference.