HPE was again positioned in the latest Gartner’s “Leaders Quadrant” For Hyper-Converged Infrastructure thanks to its newest solution – HPE SimpliVity 380 Gen 10.
Hyper-converged markets continue to grow, which is why we are aiming to popularize this newest HPE solution.
- Why SimpliVity?
- Technical specifications
Hyper-converged infrastructure is an evolution of integrated systems and the logical next step for increasing the efficiency of company operations via streamlined management capabilities, scalability, disaster recovery, backup and better storage performance. HPE SimpliVity 380 includes all of these features, distinguishing it from the competition.
Current “legacy stack” is not designed for virtualization. It consisted of 6 to 9 different devices which required significant management time, coordination, resolving software compatibility issues and patience when calling various help desks. Nutanix went a step further and incorporated server and storage system resources into their software, however omitting data backup systems, WAN optimization and caching. Use of third-party software became necessary in order to access any of these additional features. HPE SimpliVity 380 is the only optimized software platform that provides a single, shared resource pool across the entire IT stack with all of the above features and additional connectivity with cloud services. This is the only platform that has 40:1 average data efficiency and combines all IT and data-related features in a virtualization environment.
Figure 1. Legacy vs Simplivity
Core to the SimpliVity platform is the Omnistack Data Virtualization file system that uses global resource “pools” across the entire cluster. Platform also offers advanced data services:
- Guaranteed data efficiency
SimpliVity assimilates real-time deduplication, compression and optimization of all inline data from the start, once and for all.
The real strength of this platform lies in its Omnistack accelerator card that handles all heavy infrastructure operations and unburdens the x86 Intel processors which are left to do what they do best – run virtual machines. This increases the overall performance of the system by removing unnecessary IOPs. SimpliVity guarantees data efficiency of 10:1 as part of its standard warranty.
- Built-in resiliency, backup and disaster recovery
SimpliVity provides resiliency, built-in backup and “bandwidth-efficient” replication, necessary to ensure the highest level of integrity and availability of data and eliminating the need for other backup solutions.
Data protection is integrated on a virtual machine level due to such machines being easy to store, restore and transfer without worrying about data storage infrastructure.
All these options are available with only a couple of clicks within vCenter or by using orchestration tools such as Vmware vRealize Automation or UCS Director.
HPE guarantees the ability to restore a 1TB VM within one minute, meaning that the customer saved approx. 70% of time needed to restore stored virtual machines and thereby eliminated any need for using other backup or replication tools.
- Global VM-centric management
Data Virtualization Platform is “lightweight” software that makes it extremely easy to move and/or clone virtual machines between remote locations depending on application requirements. This kind of management excludes the use of LUNs, shares, and volumes, and instead is applied on a virtual machine level with a unique overview of all data centers and remote locations.
Figure 2. Omnistack DVP
Built-in data protection and automated DR
SimpliVity is the only available solution that offers a full range of protection against failures, replication as well as WAN optimization, not just for servers and storage, but the whole stack. Users can easily create backups and move VMs without having to worry about hardware infrastructure or data storage.
Figure 3. HPE Simplivity vs rival products
SimpliVity data center (cluster) consists of a series of x86 servers that manage a hypervisor such as VMware vSphere. Servers can be procured from multiple hardware vendors with common server features, such as redundant power supplies, fans, and network cards.
SimpliVity cluster is a logical grouping of such servers – combining local storage into highly available shared storage via the Data Virtualization Platform. This arrangement makes it possible for multiple SimpliVity clusters to coexist within a physical data center or a single SimpliVity data center that expands into two physical data centers, so-called metro clusters. All data center nodes are managed by a single vCenter server forming the vSphere cluster.
Figure 4. DVP
Looking at SimpliVity on a physical layer, the platform consists of SSDs connected with a RAID controller and various number of drives in the RAID field, depending on the model. Also dependent on the model is the RAID mode, with ExtraSmall and Small variants having RAID 5 drive configuration other having RAID 6.
The hypervisor can manage one or more NFS datastores containing virtual machines, and these datastores are also available to each node in the SimpliVity cluster. There are separate RAID sets for SSD drives directly linked to the OmniStack Virtual Controller. OmniStack Virtual Controller is a virtual machine and central part of the Data Virtualization Platform – NFS datastores running production virtual machines are presented to this virtual machine. OmniStack Virtual Controller ensures that local virtual machine blocks are upgraded even in remote data centers, making data storage very accessible. Physical controller is dedicated to the Omnistack Virtual Controller via the hypervisor using VMDirectPathIO technology.
Crucial part of this Data Virtualization Platform is the custom-made PCI Express Omnistack accelerator card by SimpliVity that carries out deduplication and compression and reduces processor load. Thanks to this, Omnistack Virtual Controller does not overload processor resources which can then be used to run virtual machines. Omnistack card has its own RAM, capacitors, FPGA processor and flash data storage. Omnistack accelerator card is also directly connected with the Virtual Controller using VMDirectPathIO1.
1 As shown in Figure 4
Figure 5. Omnistack DV Platform Components
Group of data centers forms a SimpliVity Federation the same way that a group of hosts forms a SimpliVity data center / cluster. SimpliVity Federation provides unified management of virtual machines across multiple data centers, operations such as backup and replication between data centers and rapid migration/transfer of virtual machines between data centers. Data Virtualization Platform spans across all SimpliVity clusters in the Federation, enabling global backup and replication of virtual machines.
Data backup policy is created and applied on a Federation level. This policy defines local and remote rules assigned to specific virtual machines or groups of virtual machines based on datastores. Topology connecting these separate data centers can be a mesh or a hub or a spoke, which will be automatically determined during Federation deployment.
Figure 6. HPE SimpliVity Federation Components
HPE SimpliVity Data Virtualization Platform
Data Virtualization Platform architecture connects the hardware and hypervisor while abstracting hardware from virtual machines. This layer also houses an accelerator card for deduplication and compression and optimizes all inline data at inception.
If we dig even deeper under the Data Virtualization Platform, we will discover two layers:
- Presentation layer – Vmware has a logical interaction with this layer
- Managing layer – layer where metadata is stored and managed
Data Virtualization Platform analyzes virtual machine logs with fine data granularity of just 8 KB. Such logs are firstly written in the accelerator card. The card simultaneously sends a synchronous reply to the second card in the second node to enable high availability (HA). When a block is written on a card, the card reports to the host. Each card has 8 GB NVRAM cache and capacitors that provide power to the card in the event of a power failure.
The card then analyzes this block, and if it determines that the block is unique, it will be compressed and written to the drive. And so on for each subsequent unique block. When a card receives a block that was already written, DVP detects this and updates the metadata of that block. It sets a pointer to that block and doesn’t write anything onto the drive.
After countless blocks pass through the DVP, Vmware thinks it has carried out dozens of IOPS’, however it has actually written a single block and updated the metadata with pointer information for that block.
SimpliVity platform protects Enterprise infrastructures in multiple ways:
- File restore – One of the most common types of data retrieval used due to unmindful or careless user actions, if an older version of specific files is required, after ransomware is detected, etc.
- VM backup – Fully independent copies of virtual machines created at a certain time that can be restored to the original virtual machine or as separate machines.
- Node – Failure protection based on a hardware RAID setup together with other components such as redundant power supplies, network cards, ECC memory, and accelerator card cache memory.
- Data center environment (cluster) – Instant VM recovery, high availability (HA), synchronous replication between hosts and RAIN
- Federation – Protection against single site failure with DR capabilities.
Simplivity virtual networking
Configuring virtual networking in the HPE SimpliVity environment is simple and flexible. There are several new connection requirements for SimpliVity Data Virtualization Platform (DVP), but beyond that, virtual and physical networking is configured the same as in any vSphere environment.
HPE SimpliVity is based on HPE DL380 Gen10 (also called SimpliVity node or host). According to technical specifications each node has two LOM network cards, 4×1 GbE and 2×10 GbE. 10 GbE card comes with a SFP + or 10Gbase-T port. Up to 3 additional network cards can be added to the system. ESXi is located on every SimpliVity host and all network configurations are done via vCenters managing SimpliVity hosts.
Guidelines/requirements or standard practice for SimpliVity DVP networking:
- SimpliVity DVP storage traffic requires 10 GbE connectivity.
- The network used for SimpliVity DVP traffic must be configured to use Jumbo Frames.
- Groups of ports associated with SimpliVity DVP Storage traffic and ESXi vmkernel have to be configured in Active/Standby mode.
- OVC Storage IP address and ESXi Storage vmkernel IP addresses have to be on the same subnet across all hosts in the cluster.
- OVC Federation IP address must be on the same subnet on all hosts in the cluster.
- SimpliVity DVP Storage, SimpliVity Federation, Management, and VM Traffic should be logically separated by VLANs.
- Network configurations should be consistent on all SimpliVity hosts in the same cluster.
- vSphere Standard vSwitch and vSphere Distributed vSwitch are supported.
- SDN, NSX are supported for VM and management traffic.
There are two possible options for physical network connectivity of SimpliVity hosts with regard to DVP traffic, 10 GbE Direct Connect and 10 GbE connectivity on switches.
10 GbE Direct Connect
When implementing a SimpliVity vSphere cluster that contains only 2 SimpliVity hosts, these hosts can be directly interconnected using 10Gb connections for enabling DVP communication between hosts (Storage, Federation). Direct Connect is supported using 10 GbE SFP+ and 10Gbase-T adapters.
When using a direct connection configuration, 4×1 GbE ports are available to handle traffic that requires less bandwidth (VM, management, etc.). These ports can be configured according to user requirements regarding traffic separation (separating management from VM traffic). Additional NICs can also be added to support additional network connectivity requirements.
10 GbE Switched
Hosts located in a vSphere cluster that contains more than 2 SimpliVity hosts (or 2 SimpliVity hosts and Standard ESXi hosts operating as compute DVP hosts) have to be connected via 10 GbE ports using s switch. All traffic in this type of implementation can be “converged” on available 10 GbE vmNICs.
Bandwidth needed for VM and management traffic depends on customer requirements. HPE SimpliVity DVP is flexible as long as DVP traffic is passing through a 10GbE link, directly or switch-connected hosts and located in the same vSphere cluster, OVC and ESXi are available to each other and can access the vCenter Server managing the vSphere cluster. Additional network adapters can be added into the configuration to provide additional network connectivity and the ability to further separate network traffic as needed.
HPE SimpliVity is a scalable platform – when faced with additional new requirements, simply add new hosts (either SimpliVity hosts or Compute hosts) to scale available resources. As the HPE SimpliVity infrastructure grows, it will be necessary to migrate from 10 GbE Direct Connect to the 10 GbE switch model. This process is very straightforward and can be done without affecting platform functionality.
Migrating from legacy stacks to SimpliVity
There are several methods for migrating virtual machines from a legacy infrastructure to SimpliVity. If the migration is planned properly, it can easily be achieved without having significant impact on VMs currently running in the virtual environment. Below is an overview of some of the more common migration methods used to move VMs from a legacy infrastructure to the SimpliVity Hyperconverged Infrastructure (HCI).
Integrating SimpliVity into an existing environment
This is probably the most common scenario. SimpliVity is used to replace the legacy environment, provide additional capacity to the current environment or provide support to a specific application. Existing vCenter is used and SimpliVity is deployed into the existing environment.
- Present a HPE SimpliVity Datastore to a host in the legacy environment.
- Use Storage vMotion to migrate VMs into the SimpliVity Datastore.
- Use vMotion to migrate VMs into a SimpliVity node.
Storage vMotion will have no impact on the availability of migrated VMs. If CPU compatibility can be established between existing hosts and SimpliVity hosts (which can be achieved by deploying SimpliVity hosts into an Enhanced vMotion Compatibility (EVC) enabled cluster), VM vMotion can be executed without affecting the availability of the virtual machine. If CPUs are not compatible, VM must be shut-down and migrated using cold vMotion. Since cold vMotion can be carried out quickly, VM downtime usually lasts only a few minutes.
If all the requirements for using Enhanced vMotion Compatibility mode are met, this function could be used for live migration of storage and compute without presenting SimpliVity storage to existing hosts and without affecting the availability of VMs during the migration process.
This migration method can also be done by presenting legacy storage to SimpliVity hosts. The process would essentially be the same but the SimpliVity design should probably include HBAs or NICs to provide SimpliVity storage connectivity with the legacy storage.
Migrating to a new SimpliVity environment
In this use case the customer runs a new vCenter environment to deploy the HPE SimpliVity platform. Typically, the reason for doing this is to get to the latest vCenter version without having to upgrade the legacy infrastructure. This is usually done when SimpliVity will completely replace the legacy infrastructure. Other reasons include moving from a Windows vCenter Deployment to a vCenter Server Appliance (VCSA), or from a physical vCenter Server to a virtual vCenter Server.
- Migrate VMs to a specific host in the legacy environment.
- Add the legacy host to the new vCenter managing the SimpliVity HCI environment.
- Present SimpliVity Datastore to the legacy host.
- Use Storage vMotion to migrate VMs to the SimpliVity Datastore.
- Use vMotion to migrate VMs to SimpliVity hosts.
In this process a legacy host is used as a “swing” host. Once the “swing” host is added to the new vCenter environment, SimpliVity Datastore is presented to this host, and it essentially becomes a compute node. The difference between using a compute node as a “swing” host and using the compute node as a permanent resource is that SimpliVity supports presenting the SimpliVity Datastore using 1 GbE connectivity for the purpose of migration – no 10 GbE switching is required.
As with the previous migration, Storage vMotion has no impact on the availability of VMs and if the CPUs are compatible then the vMotion can also be performed without any downtime. If CPUs are not compatible the VM will have to be shut-down and a cold vMotion will have to be performed.
Simplivity Federation design
HPE SimpliVity Federation design is focused on the HPE SimpliVity Arbiter’s function, Arbiter’s requirements, and the placement of the Arbiter along with some guidance on the design of the vCenter environment.
There are two key factors which influence HPE SimpliVity Federation design:
- HPE SimpliVity Arbiter requirements
- Platform Services Controller (PSC) and vCenter Server topology
The Arbiter and the vCenter Server are required components of the HPE SimpliVity Federation design. When deploying a new SimpliVity Federation, Arbiter and vCenter server have to be configured in the management network. The Arbiter can be deployed to a temporary location and then redeployed to its permanent location once the SimpliVity Federation is deployed. The vCenter environment can also be deployed to a temporary location and migrated onto the SimpliVity Cluster after deployment. There are a number of options for migrating a vCenter environment into the SimpliVity Federation.
The Arbiter is a service which acts as a witness to maintain quorum for an HPE SimpliVity Cluster to ensure data availability and data consistency should a SimpliVity host experience downtime or become inaccessible. The Arbiter service runs on a 64-bit Windows OS and requires a CPU (1 GHz), memory (1 GB), and disk drive (16 GB).
The Arbiter service can be installed on a physical server or virtual machine and there is no need to install it on a dedicated server. The Arbiter service does not use a lot of resources so it can be safely installed on a server running other services (AD, DNS, etc.). The key argument when choosing the placement of the Arbiter is it cannot be run on an HPE SimpliVity Datastore in a SimpliVity Cluster which it is witnessing.
A single Arbiter can witness all SimpliVity Clusters in a SimpliVity Federation. Multiple Arbiters can be deployed in large, multisite Federations. SimpliVity hosts communicate with the Arbiter over the management network using both UDP and TCP on port 22122. Round trip latency between the Arbiter and the SimpliVity hosts should be no more than 300 ms. All hosts within the same SimpliVity Cluster must communicate with the same Arbiter.
The Arbiter services are running in virtual machines which are hosted on the SimpliVity Data Virtualization Platform (DVP) but witnessing a different host, not the one it is running on, in the same Federation. The Arbiter in the Production Management Cluster is witnessing the SimpliVity Cluster, Recovery Management and Recovery cluster, in the Recovery Datacenter. The Arbiter in the Recovery Management Cluster is witnessing the SimpliVity Cluster, Management and Production cluster, in the Production Datacenter. This is a fully supported deployment configuration which is commonly used with larger deployments.
SimpliVity deployment has specific requirements to keep everything running smoothly:
- Appropriate physical Infrastructure (switching, cabling, tagging, ILO)
- Connection (switch or direct)
- 10gbe + 1gbe or all 10gbe10G
- LACP Enabled/Disabled
- DVS /Standard Switches
- MTU Size set for Management, Storage & Federation – 1500-Management and 9000 for Storage and Federation
- VLANs configured if using switches
- vCenter installation and configuration (creating clusters)
- vCenter version and build
- VCSA or Widows support
- vCenter Ports – 902 (UDP), 443, 80 open/Closed
- vCenter Admin account data
- vCenter Single or Linked Mode
- vCenter License
- HA and DRS turned off on the cluster
- Installation of the Deployment Manager, Arbiter and vCenter SimpliVity add-ons
- Web Client Plugin installation
- DOT NET 4.0 & JAVA 1.7 OR 1.8 (3.5.1)
- Deployment Manager installation
- SimpliVity Extension Plugin installation
- Configuring IPs across hosts
- Starting deployment
- Host1 done
- Host2 done
- Host3 done…
- System testing (HA, DRS, active-active NIC, backup, migrate, restore, storage, Federation)
How to order
HPE SimpliVity Platform is ordered using bundle configurations and there are many different bundles available with various SSD and memory sizes. Choose between 1P or 2P, 8 to 22 core CPUs. Choose between 144GB to 768GB of effective usable memory per bundle per CPU. Choose between 5 SSD models: XS, S, M, L, XL. XS and S models are configured as RAID 5 while all others are delivered in the RAID 6 configuration. You can also order two different SSD models with regard to the effective number of writes and reads: models 4000 and 6000.
Feel free to contact us for more information, testing the SimpliVity demo or free deployment education.