Virtualization Technology News and Information
White Papers
RSS
White Papers Search Results
Showing 1 - 5 of 5 white papers, page 1 of 1.
Vembu Changes the Dynamics of Data Protection for  Business Applications in a vSphere Environment
This paper examines how to use Vembu BDR to implement distributed backup and disaster recovery (DR) operations in a centrally managed data protection environment with an ingenious twist. Rather than store image backups of VMs and block-level backups of physical and VM guest host systems as a collection of backup files, Vembu BDR utilizes a document-oriented database as a backup repository, dubbed VembuHIVE, which Vembu virtualizes as a file system.
Normal 0 false false false EN-US X-NONE X-NONE In this analysis, openBench Labs assesses the performance and functionality of the Vembu Backup & Disaster Recovery (BDR) host-level (a.k.a. agentless) data protection solution in a VMware vSphere 5.5 HA Cluster. For this test they utilized a vSphere VM configured with three logical disks located on separate datastores to support an Exchange server with two mailbox databases. Each of the mailbox databases was configured to support 1,000 user accounts.

This paper provides technically savvy IT decision makers with the detailed performance and resource configuration information needed to analyze the trade-offs involved in setting up an optimal data protection and business continuity plan to support a service level agreement (SLA) with line of business (LoB) executives.

To test backup performance, they created 2,000 AD users and utilized LoadGen to generate email traffic. Each user received 120 messages and sent 20 messages over an 8-hour workday. Using this load level, we established performance baselines for a data protection using direct SAN-based agentless VM backups.

In this scenario they were able to :

  • Finish crash-consistent incremental agent-less backups in 18 minutes, while processing our base transaction load of 12 Outlook TPS.
  • Restore a fully functional VM in less than 5 minutes as a Hyper-V VM capable of sustaining an indefinite load of 4 Outlook TPS
  • Recover all user mailboxes as .pst files from a host-level agentless VM backup with no need to schedule a Windows Client backup initiated within the VM’s guest Windows OS.
In a DR scenario, Vembu leverages the ability to restore a VM in any format, to provide an Instant-boot function, When Vembu Backup Server is installed on a server that is concurrently running Hyper-V, Vembu exports the datastores associated with a VM backup as Hyper-V disks and configures a VM to boot from the datastores.
Does Backup Need a File System of its Own?
VembuHIVETM is an efficient cloud file system designed for large-scale backup and disaster recovery (BDRTM) application with support for advanced use-cases. VembuHIVETM can be thought of as a File System of File Systems with in-built version control, deduplication (elimination of redundant information to enhance storage reduction), encryption, and in-built error correction.

Backup is just not about storage. It’s the intelligence on top of storage. Typically when businesses think of backup, they see it as a simple data copy from one location to another. Traditional file systems would suffice if the need were to just copy the data. But backup is the intelligence applied on top of storage where data can be put to actual use. Imagine the ability to use backup data for staging, testing, development and preproduction deployment. Traditional file systems are not designed to meet such complex requirements.

With the advent of information technology, more and more organizations are relying on IT for running their businesses. They cannot afford to have downtime on their critical applications and need instant access to data in the event of disaster. Hence, a new type of file system is necessary to satisfy this need. 

VembuHIVETM manages the metadata smartly through its patent-pending technology, in a way that is agnostic to the file system of the backup, which is why we call VembuHIVETM, a file system of file systems. This helps the backup application to instantly associate the data in VembuHIVETM to any file system metadata, thereby allowing on-demand file or image restores in many possible file formats. The data and metadata storage, harness cluster file system and computing and storage.

This is a really powerful concept that will address some very interesting use cases not just in the backup and recovery domain but also in other domains, such as big-data analytics.

The key to the design of VembuHIVETM is its novel mechanism to capture and generate appropriate metadata and store it intelligently in a cloud infrastructure. The increment data (the changes with respect to a previous version of the same backup) are treated like versions in a version control system (CVS, GIT). This revolutionary way of data capture and metadata generation provides seamless support to a wide range of complex restore use cases.

The Next Generation Clusterless Federation Design in the Cloudistics Cloud Platform
Is there another way to manage your VMs? Yes, the introduction of the Cloudistics Cloud Platform has an innovative approach – non-clustered (or clusterless) federated design.The clusterless federation of the Cloudistics Cloud platform uses categories and tags to characterize computer nodes, migration zones, and storage groups (or blocks). With these benefits: ⦁ Node Limits Are A Thing of the Past. ⦁ Locking limitations are removed. ⦁ Flexibility is Enhanced. ⦁ Ladders of Latency are Removed. ⦁

This paper is written in the context of modern virtualized infrastructures, such as VMware or Nutanix. In such systems, a hypervisor runs on each compute node creating multiple virtual machines (VMs) per compute node. A guest OS runs inside each VM.

Data associated with each VM is stored in one or more virtual disks (vDisks). A virtual disk appears like a local disk, but can be mapped to physical storage in many ways as we will discuss.

Virtualized infrastructures use clustering to provide for non-disruptive VM migration between compute nodes, for load balancing across the nodes, for sharing storage, and for high availability and failover. Clustering is well known and has been used to build
computer systems for a long time. However, in the context of virtualized infrastructures, clustering has a number of significant limitations. Specifically, as we explain below, clusters limit scalability, decrease resource efficiency, hurt performance, reduce flexibility and impair manageability.

In this paper, we will present an innovative alternative architecture that does not have these limitations of clustering. We call our new approach clusterless federation and it is the approach used in the Cloudistics platform.

The rest of this paper is organized as follows. In Section 2, we describe the limitations of clustering and in Section 3, we drive the point home by using the specific example of VMware; other virtualized systems are similar. In Section 4, we present the clusterless federated approach and show how it avoids the limitations of clustering. We summarize in Section 5.

HyperCore-Direct: NVMe Optimized Hyperconvergence
Scale Computing’s award winning HC3 solution has long been a leader in the hyperconverged infrastructure space. Now targeting even higher performing workloads, Scale Computing is announcing HyperCore-Direct, the first hyperconverged solution to provide software defined block storage utilizing NVMe over fabrics at near bare-metal performance.
Scale Computing’s award winning HC3 solution has long been a leader in the hyperconverged infrastructure space. Now targeting even higher performing workloads, Scale Computing is announcing HyperCore-Direct, the first hyperconverged solution to provide software defined block storage utilizing NVMe over fabrics at near bare-metal performance. In this whitepaper, we will showcase the performance of a Scale HyperCore-Direct cluster which has been equipped with Intel P3700 NVMe drives, as well as a single-node HyperCore-Direct system with Intel Optane P4800X NVMe drives. Various workloads have been tested using off-the-shelf Linux and Windows virtual machine instances. The results show that HyperCore-Direct’s new NVMe optimized version of SCRIBE, the same software-defined- storage powering every HC3 cluster in production today, is able to offer the lowest latency per IO delivered to virtual machines.
HC3, SCRIBE and HyperCore Theory of Operations
This document is intended to describe the technology, concepts and operating theory behind the Scale Computing HC3 System (Hyper-converged Compute Cluster) and the HyperCore OS that powers it, including the SCRIBE (Scale Computing Reliable Independent Block Engine) storage layer.
This document is intended to describe the technology, concepts and operating theory behind the Scale Computing HC3 System (Hyper-converged Compute Cluster) and the HyperCore OS that powers it, including the SCRIBE (Scale Computing Reliable Independent Block Engine) storage layer.