Businesses of all size are increasingly starting deployments of cloud-based data, driven by the promises of greater agility, lower management cost and capital savings. It just makes sense.
When both compute and data move together – in lockstep – to the cloud, the issues to be considered are very similar to deploying or migrating an onsite application. But when the major compute operations are staying onsite and only the data is moving offsite (such as for backup, disaster recovery or compliance archive), the deployment can be more complex. In this scenario, operational executives must consider five key issues to ensure a successful experience – including meeting customer service level agreements (SLAs) and staying within budget. Based on the use case you are planning, you must consider: 1) Ongoing data transfer volume, 2) Expected frequency of usage, 3) Recall performance, 4) Application integration, 5) Ongoing management.
1. Ongoing Data Transfer Volume
It may surprise you to learn that one of the largest potential costs associated with putting data in the cloud is transport. When you buy a cloud service, you’re buying a place for the data to rest, but you need to pay to get it there. Depending on the existing network capacity you have available, the amount of data you need to transport and where the data is sourced geographically, this network impact and cost can be significant. If you can time-shift the data transfer into hours when your existing network capacity is not fully utilized, the performance and cost impact may be mitigated. So as you think about your cloud data strategy, consider not only how you will transport your initial data but also the volume of the data you will add on a monthly basis. How are you going to move this data to the cloud? Can you do the movement “off hours” to limit impact on your existing infrastructure?
2. Expected Frequency of Usage
Expected frequency (and size) of data recalls from the cloud is also a critical consideration. While the cost to leave data at rest may be as low as $0.01 per GB (or less) to store a vaulted copy and data transfer-in may be free, data transfer-out from this low performance tier can be as high as $0.10 per GB – and this is before you even consider the cost of the network transfer itself. In other words, if you create a vault in Amazon Glacier, but then end up recalling just 10% of the data you store, your bill can be up to double the size that you expected.
3. Recall Performance
This is also critical. Recall performance needs to fit within the window of your restore time SLA to your users: basically, will your cloud service provider give you back the data you are seeking rapidly enough? For example, for compliance data, you may have an SLA of hours to days, allowing you to take advantage of lower cost cold storage service options with multi-hour recall performance. But you wouldn’t want to sign up for this same service if you needed an immediate restore. Ask yourself: can recalls be planned in advance? Or do you need instant recovery? This will impact the service you select.
4. Application Integration
If your workflow environment is similar to many customers, your plan to create a data pool in the cloud will be dominated by the issue of how you integrate it with your existing applications. While flexing cloud capacity is easy, the initial setup isn’t so simple. Your existing applications are likely not set up to write to a REST (cloud) interface, nor may they be expecting the additional performance latency that a cloud may introduce. You may end up exporting data from your application and creating manual web drops or custom integration code. The workload associated with this is why many customers have decided to use cloud gateways, which bridge the gap between the remote cloud and applications used to talking local file or block device. But the decision to use a gateway introduces a new layer of technology to be selected, managed and paid for. These devices have a wide range of functionality, performance and usability. You will need to plan for this new program: selection, provisioning, application testing and maintenance. You may also wish to consider adding a layer of workflow automation as you deploy to the cloud, limiting the amount of necessary manual intervention for data migration.
5. Ongoing Management
The final issue you’ll need to think about is ongoing management. How are you going to track the usual systems management characteristics of the data stored in the cloud: capacity, performance, availability, etc.? Your cloud vendor may provide some data around this, but many leading-edge users of the cloud have found it valuable (or necessary) to also have their own cloud-centric tools to track and manage these elements, which are key to you meeting your SLAs to your customers. As with application integration, you may wish to leverage this opportunity to add a new layer of data visualization and automation to your data management.
A solid plan to manage these five considerations starts with a thorough situation analysis. For the data you plan to move, you must evaluate your data volume, practices and processes as they exist in-house today. In addition to ensuring a successful – and durable – data in the cloud architecture, this analysis is highly likely to elicit areas where you can make improvements in your onsite practices as well. You may find that – beyond giving you the opportunity to evaluate and justify new application and management automation tools – the greatest ROI in moving to the cloud will be the chance to take a fresh look at what is stored, where, and for how long. And considering how rapidly data is growing, if your pilot data-in-the-cloud project evolves into a thorough, cross-functional review of general data usage, placement (e.g. tiering) and deletion practice, that would likely be a good thing for your data storage budget – now and in the future.
These five considerations can be particularly vexing if a customer’s ecosystem is tightly integrated, as is the case with most highly demanding workflows. That’s why Quantum announced two new cloud service offerings today, Q-Cloud Archive and Q-Cloud Vault, which make it much easier to leverage the cloud in this type of environment. These solutions enable customers to move data to the cloud and store it there, all through Quantum’s StorNext data management software. The cloud basically becomes a tier managed by StorNext and fully integrated into the workflow, enabling customers to leverage both on-premise storage and cloud storage to ensure data is in the right place at the right time. This approach avoids the need for a gateway, eliminates concerns about data transfer volumes, moderates the unpredictable cost impact of data recall (by allowing for the recall of up to 5% of the data stored at no charge), provides application integration and ongoing management through StorNext, and includes end-to-end technical support from Quantum.
You can learn more about Q-Cloud Archive and Q-Cloud Vault – as well as a new cloud offering that provides a way to move data to the cloud through Quantum’s DXi backup and deduplication appliances. I look forward to hearing what you think.
Learn More About Quantum’s New Q-Cloud Solutions
In this video, Quantum’s Sr. Vice President or Product Operations and Sr. Vice President of StorNext Solutions talk about Quantum’s 3 new cloud offerings.