Controlling Cloud Costs

By keeping object data which is actively used by cloud-based applications in the cloud and storing an entire copy of data on-premises, customers can use their data center infrastructure to store a golden copy of data. This will save on metered cloud costs while continuing to extract maximum value out of their cloud services. When data stored on-prem is needed in any public cloud, customers can populate and synchronize data to the cloud location where it delivers the most business value. When the data is no longer being used in a metered public cloud, the customer can delete the data from that location without incurring egress charges.

As public cloud storage costs add up, customers will often tier their data to Glacier. While this saves money in the immediate term, if that customer wants to repatriate their data back to their active environment (S3 or primary storage), the costs are prohibitive – over $50,000 per petabyte compared to $21,000 a year to leave it in AWS Glacier. This does not include the cost of the target storage, creating a business decision around each access needed into the cold storage, asking if the cost is worth the gain of restoring that data. Vail removes the costly fees associated with large scale restores in the cloud by shifting access to local storage, removing all cloud egress and access charges associated with a traditional cloud workflow, essentially letting AWS Glacier be another disaster recovery copy of your data.

Data Flow vs Metadata Flow

One major benefit of Vail is that data does not flow into and out of the public cloud unless a cloud bucket is specified in the policy as a storage target. If a Vail policy dictates that copies are to be stored and accessible at two different physical sites, then the data movers in each Vail node transfer data directly between those sites and not through the public cloud. This prevents excessive egress charges from cloud vendors and removes an additional network hop.

If a policy requires a copy on a cloud bucket as well, data is always copied directly between nodes and to the cloud bucket. A cloud bucket is never used as a source, and that is the only copy of data available.

1. Metadata is always stored and controlled in the cloud. Transactions require multiple interactions with the cloud Management Server including metadata access, location recording, loading of the job queue, etc. Database transactions, however, are low bandwidth, low cost, and can be fully encrypted. Metadata is also locally cached to maximize performance.
2. Through this architecture, Vail delivers a single management view into all data regardless of its physical location. Data is transferred from site to site, saving on egress fees and cloud storage charges. The transactional, low-capacity metadata is stored and accessed from the public cloud, saving up to 80% over a traditional pure cloud workflow.

Vail Use Case – Government Agency

The Goal:
To find a way to collect globally dispersed data into a central repository where a supercomputer can perform analytics and generate results. Once results are generated, they must be shared with other departments and users.

The Challenge:
Remote offices and remote collection sites were unable to connect to a central location to transfer data, causing delays and other challenges of transferring data. Once results were generated, the public cloud was used to distribute the data. This resulted in unplanned egress and out-charges with the public cloud and alternatives needed to be evaluate.

The Solution:
With Vail, this government agency was able to implement an end-to-end data management and delivery service that allows data collection from instruments and contributors from all over the world using a combination of edge, core and cloud computing locations. In addition, they can store data in AWS and on on-premises according to a policy defined to meet service levels that match the value of the data over time, while managing access costs to adhere to the needs of the data consumers and their budget resources.

Lastly, Vail is ab le to keep historic data in a multi-site, durable, accessible, and affordable datastore. Any dataset can be accessed in minutes while only paying storage costs.

With Vail, this agency is able to leverage any combination of public clouds and on-on-premises datastores. They are able to choose the storage tier that matches the value of their data and response time needed for any point in time access to the data. This happens while managing the data based on the elasticity and reliability of AWS cloud services.

Learn More >

*Amazon Glacier is a registered trademark of Amazon Technologies, Inc.

This short fun animated video shows how Spectra’s Vail, a distributed multi-cloud software, unifies data to enable on-premises applications and cloud services — no matter where data is created or stored.

Watch how Vail transforms workflows

Interested in a live Vail demo?

Fill in the form and see what Vail can do for you.


Spectra Logic
Follow Us