• Skip to main content
  • 855-855-8300
  • Contact
  • Architecture
    • Overview
    • Benefits
    • Schedule a Demo
  • Use Cases
    • VMware Alternative
    • VMware DR
    • SAN Replacement
    • Hyperconverged
    • Small Data Center
    • Secure Research Computing
  • Resources
    • About VergeIO
    • Blog
    • Documents
    • In The News
    • Press Releases
    • Support
    • Technical Wiki
    • Videos
    • Webinars
  • How to Buy
    • Versions
    • Test Drive
×
  • Architecture
    • Overview
    • Benefits
    • Schedule a Demo
  • Use Cases
    • VMware Alternative
    • VMware DR
    • SAN Replacement
    • Hyperconverged
    • Small Data Center
    • Secure Research Computing
  • Resources
    • About VergeIO
    • Blog
    • Documents
    • In The News
    • Press Releases
    • Support
    • Technical Wiki
    • Videos
    • Webinars
  • How to Buy
    • Versions
    • Test Drive

Storage

Snapshots or Clones for Data Protection?

April 25, 2023 by George Crump Leave a Comment

snapshots or clones for data protection

Most storage solutions will provide IT professionals with either snapshots or clones for data protection, but are the differences between the two functions significant enough to make it part of your selection criteria? Like all things in IT, the answer depends.  In this case, it depends on if and how your vendor implemented the two technologies. 

Register now to join us live on May 4th for technical deep dive into virtual infrastructure file systems and see a live demonstration of IOprotect.

What Are Snapshots?

Deciding if snapshots or clones are best for data protection first requires understanding how the two technologies work. First, let’s look at snapshots. Most storage solutions, be they a filesystem or block storage, have a metadata layer that points to where each data segment resides. A snapshot makes a copy of those pointers at a specific time and then sets those blocks pointing to a read-only mode until it expires. 

Snapshot Update Methods

There are two methods for updating a read-only segment because of a snapshot.  The first method is a copy-on-write process. When a user or application attempts to update or change an existing segment, the storage solution copies the old segment to a new location and allows the new data to occupy the original segment. The storage solution then updates the snapshot metadata with the old segment’s new location. 

The second snapshot update method is “redirect on write”. Using this method, the storage system will write the modified data to a new location and update the metadata of the “production view” of the data. It does not need to update the “snapshot view” of the data. 

Both of these methods limit the scalability of snapshots because multiple writes and multiple changes to metadata need to occur. Also, many storage systems use separate metadata trees to manage each snapshot. As the number of snapshots increases and the depth of those snapshots (snapshots of snapshots), the complexity of managing and updating the metadata wears on system performance.  As a result, where the snapshot is occurring within the hypervisor, on the same hardware as the hypervisor (software-defined storage running as a virtual machine), or on dedicated storage hardware, there are limits to how many snapshots the storage solution can maintain. 

The complexity shows itself by degrading overall system performance. Storage systems with legacy snapshot technology require:

  • Limitations to number of copies retained
  • High-end processors in the storage servers
  • Dedicated data processing units (DPUs)
  • Days to remove old snapshots

What Are Clones?

Clones are copies of existing segments. They are more standalone, and updating a clone does not require the same metadata overhead as snapshots. The independence of a clone means that they don’t suffer from performance degradation as snapshots regardless of how many there are or how long they are retained. Clones don’t need either of the sophisticated update methods that snapshots require.

The Downside to Clones

The typical downside to clones is that they are either complete copies of the original volume or deduplicated copies. A full copy, means that data must traverse the internals of the storage infrastructure, travel across the network to the hypervisor, and back down the network again to the storage system. 

Some hypervisors have initiated capabilities to eliminate traversing the network, saving time. Still, most cloning functions must process data through the internals of the storage solution twice, even if that solution has a deduplication feature. With deduplication, the resulting clone may not consume any additional capacity, but the time to create that copy is still significant, especially if the volume is of any measurable size. It is best not to use the applications while the storage solution clones it. As a result, most organizations don’t use cloning as part of their data protection strategy. 

IOclone — The Best of Clones and Snapshots

As we’ve discussed, clones and snapshots typically have some overhead in the capacity they consume and the processing required to use them. Clones typically have to make a copy of all of the metadata information, which means the cloning process takes some time upfront, but then they are ready to use and independent. Snapshots don’t have the upfront processing time and, as a result, are ready for use almost instantly. However, they show performance degradation as the number of snapshots increases when used or during snapshot clean-up routines. 

IOclone is a capability of the VergeOS operating system that combines the best of clones and snapshots into a single solution. Since global inline deduplication is part of the metadata in VergeOS, IOclone, copies are similar to snapshots. Regardless of capacity, it can create clones of VMs, volumes, or entire virtual data centers in milliseconds. At the same time, IOclone-created copies have the stand-alone performance of independent clones without initially consuming additional capacity footprint.

With IOclone, IT doesn’t have to choose between snapshots or clones for data protection. This capability within VergeOS can retain hundreds, even thousands of copies of VMs, volumes, or entire Virtual Data Centers (VDC) without negatively impacting performance or capacity consumption.

Learn More

  • Register for our live TechTalk: Deep Dive on Virtual Infrastructure File Systems
  • Subscribe to our eBook: “Designing a Resilient Infrastructure“
  • Review our IOclone Datasheet

snapshots or clones for data protection

Conclusion

IOclone is also part of our IOprotect solution, which enables you to start a VMware Exit by first using VergeOS as a disaster recovery solution. Most customers find IOprotect reduces the cost of disaster recovery by more than 50% without adding additional hardware. It provides a complete recovery environment, converging disaster recovery so that data, applications, and the processing power to recover are all available from a small cluster of nodes. 

As your confidence in VergeOS grows, you can use it for your production environment. The tightly integrated VergeOS architecture delivers more efficient performance, increasing workload density on less physical hardware. Once your conversion is complete, you’ll lower costs by as much as 80% and enjoy an actively developed data center operating system with unparalleled support.

snapshots or clones for data protection

Filed Under: Blog, Storage Tagged With: dataprotection

The Total Cost of Replacing a SAN

January 2, 2023 by Verge.IO Leave a Comment

Storage vendors force you to replace your storage system every four to five years, so understanding the total cost of SAN replacement is critical for IT professionals to learn. After all, replacing a SAN (storage area network) or NAS (network attached storage) has more expenses than just buying the new hardware.  

There are many reasons for replacing a SAN or a NAS. At the top of the list are artificial end-of-life, inability to meet new performance or capacity demands, lack of flexibility to support new hardware innovations, and inadequate protection from new, modern threats like ransomware. The result is a never-ending storage refresh merry-go-round. 

There are five components to understanding the total cost of replacing a SAN:

‍

1. The Cost of SAN Identification 

The first element of the total cost of replacing a SAN is identifying a set of potential storage replacement candidates. The IT team has to play the role of storage industry analyst to find the right storage system to meet their current and future needs. When done correctly, this process can take weeks; when done hastily, this process can lead to the purchase of a storage system that doesn’t quite meet the current requirements or is more expensive than necessary. Since most IT staffs are shorthanded and stretched too thin, the chances of not selecting the right SAN or NAS replacement candidates are high. 

‍

2. The Cost of SAN Proof of Concepts 

After making a short list of potential replacement storage systems, IT must evaluate and test each manufacturer’s claims. Creating a test environment that can simulate a real-world workload has always been challenging; now, supply chain issues and staff shortages make testing almost impossible. Most vendors can’t just send you software and let you install their storage solution on available hardware. Instead, they must ship you pre-configured hardware and be heavily involved in your evaluation’s proof of concept phase. It can take months just to get the storage system so you can test it. You also have to ask yourself, “if I can’t install it myself, how hard will it be to support and operate it after they leave?”

‍

3. The System Cost

While all vendors claim an impressive return on investment (ROI), they will not give you the system for free. You need to pay for it and hope it lasts long enough that your organization will realize the ROI. There are two keys to being able to realize a solid ROI. First, the upfront cost must be low enough that the organization can afford the system. Second, the operational costs can’t be too burdensome, and third, the system has to stay in operation long enough for the ROI to be realized.

‍

4. The Cost of SAN Migration 

Once the new storage system is selected and IT completes its evaluation, it must migrate data to the new system. The larger the environment, the more challenging the migration process becomes. In most cases, IT must identify and purchase special software tools to help with the migration. The average time to migrate from an old storage system to a new one is more than 90 days. In most cases, it is more than six months. Both the old and the new storage systems must stay active during this time (which doubles power and cooling costs), increasing backup complexity and is a cause of many operational errors. By the time the migration is complete, IT is only three years away from having to start the process all over again.

‍

5. The Cost of Learning a New System

Once data is migrated to the new storage system. IT administrators must learn the new system’s software. There will be new methods to create volumes, configure drive redundancy, take and retain snapshots, and implement replication. There are often some features missing on the new system, causing IT to create new workarounds. 

‍

The Problem with “Stretching” The Current System 

To avoid the storage refresh ordeal, many organizations will try to stretch the serviceable life of their current storage system. If the system continues to meet performance or capacity demands, this stretch typically involves buying extra out-of-warranty maintenance from the manufacturer or a third party. These out-of-warranty contracts are costly and more challenging to support.

There is also a stretching cost of lost capabilities because the storage system can’t provide improved performance or increased capacity as soon as users or applications need it. Lack of performance or capacity has a ripple effect on the networking and compute tiers, which increases the cost of those infrastructures. At some point, stretching the storage system leads to a break in data access or response time, leading to scrambling through the storage refresh process.

‍

Storage Silos and Three-Tier Architectures Amplify the Costs 

One of the biggest challenges with storage replacement is the fragmentation within the data center. The data center is islands of separate workloads. Each of these islands typically has its own computing infrastructure, sometimes its own network infrastructure, and almost certainly its own storage system. While a storage system typically requires refresh or replacement every four to five years, the separate storage islands mean that most organizations have six or seven different storage systems. As a result, IT is almost continuously replacing storage systems and managing data migration processes.

‍

How to Get off the Storage Refresh Merry-Go-Round 

Getting off the merry-go-round requires consolidating infrastructure so that IT has only one entity to manage instead of six or seven. It means a storage infrastructure that’s flexible enough to meet today’s and tomorrow’s needs and support both legacy and future storage hardware. The new solution can’t compromise on features. It has to provide a robust set of data services that enable organizations to protect their data assets and assure their availability. 

‍

VergeIO’s VergeOS eliminates complicated data center processes like storage refreshes. We simplify IT and end storage refreshes by creating an ultraconverged infrastructure that collapses the data center tiers (compute, network, storage) into a unified, cohesive operating environment.

‍

‍

‍

Next Steps

  • Learn more about VergeIO’s storage capabilities (don’t compromise data)
  • Register for a free test drive (don’t wait months for a POC)
  • Watch our On-Demand Webinar “Three Infrastructure Questions IT Leaders Must Answer in 2023”

Filed Under: Storage

855-855-8300

Get Started

  • Versions
  • Request Tour
  • Test Drive

VergeIO For

  • VMware Alternative
  • SAN Replacement
  • Hyperconverged
  • Server Room
  • Secure Research Computing

Product

  • Benefits
  • Documents
  • Architecture Overview
  • Use Cases
  • Videos

Company

  • About
  • Blog
  • Technical Wiki
  • License Agreement
  • Terms & Conditions

© 2023 Verge.io. All Rights Reserved.