RackWare Best in Class Solution
- 7 hours ago
- 4 min read
Why Filesystem-Level Replication Outperforms Storage Sector and VMDK Approaches
RackWare employs a unique, patented, delta sync mechanism to execute Migration, DR, and backup operations that has the largest scope of use cases in the industry. It is a Filesystem mechanism, operating on used data, not a storage sector or VMDK based approach.
Products that employ a storage sector or VMDK approach often have issues surrounding data consistency that prevents applications and servers from booting and initializing properly. Manual intervention, and time, are required to restore the applications to their functioning state. The same issues can also affect groups of applications booting on different servers and hypervisors and successfully initializing them to operational state.
The Issue of Data Consistency and Application Recovery
Except for shared distributed databases, applications rarely, if ever, coordinate writes to disk as doing so requires complex protocols. Those protocols would be especially complex when involving more than two servers. Without those protocols, server IOs to disk are extremely non-deterministic across servers. OSes (except for distributed databases) do not coordinate writing IOs with each other; they schedule and execute them based on their own workload dynamics. IO writes can be delayed by higher priority processes being invoked, interrupts, paging and many other factors.
What is highly deterministic, and what applications rely on, is that the IOs from an application are done not only in order but in complete groups. A transaction may consist of numerous IOs over the time it takes for the entire transaction to be processed. With applications increasingly using in-memory staging and caching, IOs are flushed very non-deterministically unless forced by the application itself or Operating System mechanisms.
One of the core problems that VMDK solutions have is that, because of where they perform the replication, they only see the writes directly to disk. Prior to higher memory servers and applications not utilizing memory staging and caching, this was not as much of a problem. But in modern systems and applications this has highlighted a fundamental design flaw in VMDK solutions.
Over time products have added complex features to address this data consistency problem. Initially, improvements such as Journaling were touted as features but they were really a way to disguise and circumvent the design flaw of IO inconsistency. Journaling helped but more invasive features were required to correctly remediate the issue, usually involving Operating System hooks to force the flushing of IOs. Appended features on top of a fundamentally different architecture creates complexity with the result that these solutions rarely work properly, bolting on an unsightly addition that is inconsistent with the core workflow of the base technology.
RackWare is designed for modern systems, applications, and Cloud environments, effectively avoiding this and other fundamental design flaws. The RackWare solution is inherently data consistent because of the unique way delta syncs are done. Its replication and sync technologies are integrated with standard Operating System mechanisms to coordinate the correct flushing of IOs to disk via OS level, logical volume snapshots. It then replicates and syncs from a static, consistent version of the data. No special configurations are required; this happens automatically and reliably. Less labor is necessary for both setup and, more importantly, when restoring platforms and applications in a Target environment. RackWare has been used on over 1 million physical servers and Virtual Machines across numerous Hypervisors and all major public and private Clouds, ensuring consistent, immediate recovery.
Aspect | Storage Sector / VMDK Approaches | RackWare Filesystem-Level Replication |
Replication method | Storage sector or VMDK level | Filesystem level, operating on used data |
Visibility into writes | Only writes that reach disk | OS-level logical volume snapshot captures consistent state |
Data consistency | Compensated via Journaling and OS hooks | Â Inherently data-consistent by design |
Application recovery | Often requires excessive manual intervention to bring applications to functional state | Complete restoration of application functionality |
Network outages | Sensitive, often requiring complete re-replication | Handles even long-term outages without re-replication |
In-memory applications (e.g. SAP HANA) | Requires special configuration or product extensions | Standard operation, no extra configuration |
Initial replication scope | Entire disk | Only used data |
Disk defragmentation | Sensitive | Not sensitive |
Other RackWare Benefits
Importantly, RackWare's unique solution is surrounded by equally superior technologies, often overlooked, to form a complete solution in the mobility and resilience market.
First is a Discovery and Assessment function that quickly and accurately ascertains information about the scope and potential costs of the project. It is unique in that it’s fast and effective, avoiding unnecessary, redundant information that complicates the Assessment. Multiple views are provided enabling the development of an overall plan from inception to steady state with timelines, costs, and logistics.
Another differentiator is that RackWare provides deployment tradeoffs between cost and performance. Different classes of servers have different mobility and resilience needs. Unlike other solutions, RackWare has the options to meet those differing needs in a single easy to use solution. Lower cost deployment options are likely applicable for the majority of servers and higher cost deployment options can be implemented to meet the strict performance requirements of any Enterprise application.
The Policy Engine also provides unique flexibility and granularity. It finds a balance between functional options and ease of use.
RackWare provides tracking and report generation for all operations including DR drills and cutover operations that can be used to meet audit requirements for both Public Sector and standards across multiple industries.
The RackWare proprietary replication and sync engine is designed to ensure data consistency while optimizing data transfer and maintaining strict security standards. The replicated server is automatically provisioned and boots in any target environment with networks automatically configured and applications operational.
All these technologies create a comprehensive set of advantages over traditional sector level and VMDK based products.
Selective sync at the drive, directory and file level
Highly flexible DR policies with granularity of frequency and selected data
Converged backup features such as retention policies and single file restore
Far superior data consistency ensuring applications achieve production operational state with little to no intervention
Permits DR test without bringing up a second site
Not sensitive to network outages, never requires a complete re-replication of data no matter how long the outage
Flexible and better storage options on target side
Easy automation for both keeping IPs and updating IPs
Supports in-memory applications (e.g. – SAP HANA) as part of standard operations with no special or additional configuration or product extensions
Initial replication is more efficient as it only replicates used data as opposed to the entire disk
Not sensitive to disk defragmentation
Single file restore and restore anywhere
BIOS → UEFI and UEFI → BIOS conversion