ZebraHost Hyper-Converged Appliance, HCI, HCA, Hyper Convergence

Q. What is a Gridstore HyperConverged Appliance (HCA)?

A. A Gridstore HCA is the building block of the Gridstore HCI solution that delivers all-flash, bare-metal performance. Each HCA combines compute, storage area network (SAN) and storage into a single tier, allowing true fault tolerance with efficient scaling.
A four-node Gridstore HCA enclosure can have:

  • All-Flash HyperConverged Nodes (FCNs)
  • All-Flash Storage-only Nodes (FSNs)
  • A mixture of FCNs and FSNs
  • Up to six drives per node

A two-node Gridstore HCA enclosure, for graphics-intensive workloads, can have:

  • All-Flash Graphics Nodes (FGNs)
  • A mixture of FGNs, FCNs, and FSNs
  • Up to 12 drives per node

Unique to HCI solutions, the Gridstore all-flash HCI incorporates a scale-to-fit design that allows storage growth separate from compute growth, eliminating costly compute sprawl. Single-pane-of-glass management, which can integrate with Microsoft System Center, ensures lower operating expenditures (OpEx) while true Quality of Service (QoS) ensures the right data is available to the right virtual machine (VM) when it is needed.

Q. What are the benefits of Gridstore HCI?

A. The Gridstore HCI solution eliminates the separate layers of compute, storage and SAN and converges them into a single tier, reducing cost and complexity. This eliminates 50% of up front equipment cost and 75% of the ongoing operational costs, including data center power and cooling, as well as the licensing stack for hypervisors, management software, and maintenance. The high performance all-flash architecture supports higher VM density found in HCI deployments, while inline erasure codes lead to even higher VM density.

Q. What problems does Gridstore solve?

A. Gridstore solves three key problems in the data center to decrease TCO and increase business agility:

  1. High cost of replication
  2. Loss of VM storage optimization
  3. Efficient infrastructure scaling

High cost of replication: The three-way replication architecture of first-generation HCI is considered standard in Webscale architectures using commodity hardware. While this model delivers the required protection, it comes at a high cost. When a write occurs on one node, the full write is replicated to two other nodes in the cluster, generating the following costs:

  1. It creates high latency due to the synchronous replications that must complete before the I/O is committed
  2. It doubles network traffic to achieve this—for every write, two replicas are sent across the network to two other hosts
  3. It consumes 3X the storage resources—for every TB written, three TBs are consumed
  4. It wastes 50% of I/O operations per second (IOPS) available in the system—assuming a 75/25 read/write split, the replica writes consume 50% of the available IOPS

Loss of VM storage optimization: Traditional storage was designed to present a logical unit number (LUN) to a server. This worked well until that server became a hypervisor running 10, 20 or 30 virtual servers, all pointing to the same LUN. Adding to the complexity, all hypervisors need to point to the same LUN to enable VM mobility or high availability. Architecturally, this creates a many-to-one relationship that makes optimization impossible. Each VM has a different I/O pattern. When they all mix together, the aggregate I/O becomes highly random, creating performance issues with no way to isolate problems or control I/O per VM.
Efficient infrastructure scaling: One of the major drawbacks of first generation HCI is that it forces users to scale compute with storage. Storage is consumed daily and typically grows at 4-6 times the rate of compute. When they consolidated compute and storage into a first generation hyper-converged system, customers were forced to grow the entire infrastructure at the rate of storage. This lead to a massive compute and hypervisor cost and sprawl that adds to management cost and complexity.

Q. How does the Gridstore HCI solution solve these problems?

A. To solve the problem of high cost of replication (a), Gridstore HCI is designed to operate in the kernel at the source where it uses erasure encoding to protect the data before it leaves the host. Erasure encoding is an advanced forward error correction method used for communications. Erasure encoding breaks a block of I/O into N fragments and encodes these fragments. This means that a number (K) of these fragments can be lost but the data still can be reassembled as long as (N-K) fragments are available. The encoded fragments are distributed in parallel to N resource controllers anywhere in the Grid. This method protects against multiple component failures, communications failures, and entire node failures.

To solve the problem of loss of VM storage optimization (b), Gridstore HCI is designed with a core component, the vController™, which operates in the host kernel and appears as alocal SCSI device. From this privileged position, it isolates I/O from each of the VMs into independent swim lanes or channels from the hypervisor across the network and onto storage nodes in the Grid. This recreates the 1:1 relationship between VM and storage. Once I/O is isolated, each channel can be independently optimized according to what the VM is trying to achieve, while happening dynamically to optimize the I/O pattern to the demands of the VM.

To solve the problem of efficient infrastructure scaling (c), Gridstore HCI decouples storage from compute, which means as you add more storage, you don’t need to add more compete power. Gridstore runs bare metal which in this context means that you do not require a hypervisor to run on, allowing you to put Gridstore software on any hardware appliance and have it add that storage into the pool. There is no requirement for CPU or RAM to run a full hypervisor or an NFS host requiring 32GB.

Q. What features in Windows Server 2012 R2 are supported?

A. Gridstore leverages its strategic partnership with Microsoft to support features, such as:

  • Deduplication
  • VM Replication
  • VM Snapshot (Checkpoints)
  • Azure Site Recovery
  • Cluster Support
  • Seamless Management per VM
  • Storage QoS
  • Microsoft Hyper-V
  • SMB3

Q. What are the configuration options available with the Gridstore solution?

A. Meeting the performance and storage needs of Windows workloads is simplified with Gridstore HCAs, which include compute, storage, and storage networking all in a single system. Starting with as few as three nodes and scaling up to 256 with over 6PB of all-flash capacity, Windows workloads are supported with the power and storage that fits an IT budget, while allowing easy scaling up as needs change. Choose from these three options to achieve the most efficient, high performance, cost effective infrastructure with lowered upfront costs, ongoing lowered CapEx and OpEx, and VM optimization:

  • HCA with dual processor E5-2690 v3, 32GB – 1TB DRAM (configurable) and 2x10GbE RJ45 ports or 10G SFP+
  • HCA with dual processor E5-2640 v3, 32GB – 1TB DRAM (configurable) and 2x10GbE RJ45 ports or 10G SFP+
  • HCA with dual processor E5-2620 v3, 32GB – 1TB DRAM (configurable) and 2x10GbE RJ45 ports or 10G SFP+

Selecting the right storage is also a key requirement for supporting Windows workloads. Gridstore offers three SSD options:

  • 960GB SATA III 2.5” SSD for mixed workloads
  • 1920GB SATA III 2.5” SSD for read-intensive workloads
  • 3840GB SATA III 2.5” SSD for read-intensive workloads

Q. Does the Gridstore solution lock me into a proprietary solution with components that can’t be used for other purposes?

A. No, in fact, the Gridstore architecture allows you to place our software on your existing Microsoft servers, so they can access that same high performance flash tier of storage. We not only allow you to scale compute and storage independently, we also allow you to fully utilize your existing infrastructure investments in the same grid. Legacy infrastructure (both servers and storage) can continue to operate as part of the Gridstore solution. Unlike our competitors, our goal is to NOT create another silo. Our vController (software driver that provides a SCSI block device to the host) can sit on any existing Windows Server or host. These hosts can then have access to all-flash storage in the pool the same as our HyperConverged Nodes. All of this infrastructure can then be managed through a single management console, such as Microsoft Systems Center.

Q. Where is the Gridstore HCI solution most used?

A. The Gridstore HCI solution is ideal for infrastructure refreshes, private cloud deployments, and mixed and specific Windows workloads, such as VDI, SQL Server, Dev/Test, and remote office/branch office (ROBO).

Q. Does the Gridstore storage support CIFS volumes?

A. Yes, we provide SCSI block devices to the hosts. The host can mount and put any file system it wants on this. The host then provides the CIFS/SMB services to its clients.

Q. Does the Gridstore VDI hyper-­converged solution give the option for NVidia add‐on graphics?

A. Yes, there is one available PCIe slot per server node. For the specific NVidia GRID cards, they are supported both by our hardware and the Microsoft or Citrix virtualization stacks.

Q. Is this all OS-agnostic or does it require a specific version of Linux or Windows? Will it allow instances of Mac OS-X?

A. The Gridstore HCI solution runs on Windows Server / Hyper-V. Hyper-V can run any OS you require in a VM.

Q. Can a user run native Windows on the nodes or only VMs in Hyper‐V?

A. The Gridstore HCI solution runs on any host that runs Windows Server 2008R2 or later. Hosts can be either physical hosts, for example running SQL Server or Exchange, or they can be virtualized hosts running anything supported by the hypervisor, including Linux.

Q. Can we run SQL /File clusters?

A. Yes, we run bare metal on the server and offer excellent performance for SQL Server applications typically with around 1ms latency for small I/Os.

Q. What is the performance and are the numbers for SLC or MLC?

A. Our performance numbers are for eMLC and are guaranteed for 3.6 Full Drive write per day for five years. This is for 100% Random 4K writes. For example, for a 1TB drive, you can write 3.6TB per day for five years or a total of 6.5PB. For sequential I/O, this goes up to 22 full drive writes per day for five years. Note that these numbers do NOT assume any data reduction, such as dedupe or compression.

Q. Do you need to start with predictable performance for SQL Server or will it scale as needed?

A. You can scale incrementally either compute or storage as your needs grow. The Gridstore HCI solution is excellent at SQL Server consolidation, Business Intelligence (BI), data warehousing, etc. We are working with Microsoft to develop a complete BI in a box, which you can then scale incrementally as your environment requires.

©2015 ZebraHost LLC | All Rights Reserved | Powered by ZebraHost