Not All Workloads Are Created Equal
Not all application workloads are created equal. They each have their own requirements and business value. They need different resources, and have varying levels of importance and impact to your organization. This is an important consideration when you’re planning data protection and disaster recovery (DR).
For the purpose of data protection, you could treat all your workloads the same and apply identical policies. But a “one size fits all approach” isn’t the most effective or economical. Either you’ll end up spending more than you want to apply the highest standard of protection to everything, or you’ll end up with less coverage than is ideal for your most important workloads. A more nuanced approach is required.
To find the best data protection strategy for individual workloads, consider the role of each and its importance to your organization. Also take into account the resource requirements of different workload types. Then think about which part of the data protection puzzle will give you the best coverage for the lowest cost.
Data protection can be roughly divided into three different strategies: disaster recovery, backups, and archives. Disaster recovery involves keeping two identical copies of a workload running in lockstep. If there’s a problem with the production copy of a workload, the workload fails over to the disaster recovery infrastructure as quickly and seamlessly as possible.
Disaster recovery’s primary concerns are Recovery Time Objectives (RTOs) and Recovery Point Objectives (RPOs). RTOs are a measure of the time it takes to get a workload back online, while RPOs are a measure of how old the data in a backup or disaster recovery copy is.
The key question for determining your organization’s RTO tolerance is: “How long can this workload be offline before there’s a problem?” While it would be ideal to frame all disaster recovery questions as simply, “How quickly can I get this environment back up and running?”, not all applications are equal, budgets are not infinite, and disaster recovery planning usually starts by asking, “Which are the applications that you cannot function without?”
Conversely, which applications could be unavailable for several hours without anyone being terribly irritated? Which applications only get used every few days? These questions determine which workloads must use which type of disaster recovery product to provide high availability in case of an outage.
Next, consider your backup strategy. Copies of your data need to exist in at least two places. This is essential in the case of a ransomware attack or a catastrophic human error that deletes a vital database. Some workloads should be backed up more often than others—these are workloads that see frequent changes. Your CRM software may need to be backed up multiple times per day, but a web server for a mostly static web page could be backed up less frequently.
This is the part of data protection where RPOs are discussed. The key question for RPOs is: “How much data can you afford to lose?” How much would be lost if you needed to restore from an image taken three hours ago versus three days ago? Of course, an RPO of near-zero sounds attractive. But you can achieve this with different backup schedules for different workloads.
The data that you use—and change—on a daily basis needs to be imaged frequently so that it can be brought back with as little loss as possible in the event of a failure. But the files from last year’s marketing campaign are unlikely to change often, and can be put on a less frequent backup schedule.
The consideration here is whether your workload involves a lot of writes (changes) or is mostly read-only. Deciding on backup schedules that make sense can save you money while still protecting your most important digital assets.
Backups and disaster recovery are closely intertwined. RTO and RPO questions and calculations are not entirely inseparable, and discussions about requirements for one often inform requirements for the other. Consider, for example, SAP HANA.
For some very high-transaction workloads, mirroring is your best strategy for data protection. If you’re running a HANA database, for example, regular backups alone won’t be adequate, and many types of replication-based data protection solutions will prove inadequate. Even if you take backup snapshots frequently, there might be thousands of changes between one snapshot and the next. Restoring from a backup could mean unacceptable data loss, so you need to avoid downtime altogether.
However, you probably don’t want to replicate every single workload in this manner. For one thing, it will be expensive. Also, network connectivity may place a hard limit on how much data you can keep in sync. Even if you pay for high bandwidth, in some areas there are only a few backbone cables serving the entire region. Data going in or out competes for space with the traffic from everyone else in the region. In this case, it may be useful for you to have your mirrored workloads hosted elsewhere, to avoid contention.
The final piece of the puzzle is archiving. Some data must be kept for legal reasons, but will seldom be needed—perhaps only in the event of an audit. In order to decide which data to archive, ask yourself this question: “Will I ever look at this again, really?” If the answer is “no,” then this data doesn’t need to take up resources in your production data center. But you do need to be able to get it back within a reasonable time if requested.
Tapes used to be the primary archive solution, and are still used by some organizations. However, tapes can be difficult to restore from. A better solution might be to put your archive data in low-cost cloud storage. Then if it needs to come back, it can be “summoned” with a click.
iland has solutions to help with each piece of the data protection puzzle. iland offers disaster recovery to meet a variety of needs. iland’s award-winning VMware-based DR environment is solid and secure. iland partners with many data protection vendors, including Veeam, to provide both backups and disaster recovery approaches that can meet any combination of RTOs and RPOs that might be required.
iland complements all backup and DR products with best-in-class onboarding, integration, and support. iland isn’t a one-size-fits-all solution; it’s many solutions custom tailored to fit you.
[su_box title=”Sponsored Article” box_color=”#0aa3e3″ radius=”4″]
|This article is sponsored by iland.
iland is a global cloud service provider of secure and compliant hosting for infrastructure (IaaS), disaster recovery (DRaaS), and backup as a service (BaaS). They are recognized by industry analysts as a leader in disaster recovery. The award-winning iland Secure Cloud Console natively combines deep layered security, predictive analytics, and compliance to deliver unmatched visibility and ease of management for all of iland’s cloud services. With headquarters in Houston, Texas, London, UK, and Sydney, Australia, iland delivers cloud services from its data centers throughout North America, Europe, Australia, and Asia.