HP Helion Rack - a solution for the rapid deployment of a private cloud based on OpenStack
In the first article of this series, we talked about the HP Helion product family, which significantly improves the use of OpenStack cloud technologies in the corporate sector, and now let's move on to describing the latest HP Helion news, which was introduced in late March.
HP Helion Rack is a fully configured, configured and tested platform for building a private cloud based on OpenStack, designed for the design, development and deployment of cloud applications. The main purpose of HP Helion Rack is to host applications that require a reliable server infrastructure with optimally tuned performance. It uses HP Helion OpenStack software (augmented by HP’s OpenStack Distribution Module) and HP ProLiant DL360 / DL380 Gen9 ninth-generation rack servers installed in the same rack and integrated using HP FlexFabric 5700 switches. In fact, HP Helion Rack is a “turnkey solution” »For the rapid implementation of the OpenStack cloud, eliminating the need for the customer to engage in the design of the frequent cloud, installing and configuring the OpenStack package, deploying and configuring the cloud-servicing server equipment.
HP considers the development and deployment of applications for use in the cloud (cloud native), the rapid deployment of infrastructure for developing and testing applications, and acting as a private cloud resource broker for “heavy” cloud applications as the main use cases for HP Helion Rack.
')
The first image shows the main software and hardware components of the HP Helion Rack — the HP Helion Development Platform application platform for providing PaaS services, the HP Helion OpenStack cloud deployment platform, and the HP ProLiant Gen9 servers.
HP Helion Rack supports various scaling options depending on the target destination, so the specific hardware configuration of the servers and their number depends on the tasks of the cloud that will be deployed on them. As cloud controllers and computing nodes, the solution uses HP ProLiant DL360 Gen9 single-duplex dual-socket servers, and HP ProLiant DL380 Gen9 dual-dual two-socket servers as storage nodes, where you can install up to 12 full-size disks.
HP Helion Rack Architecture
The minimum configuration of this solution consists of the following servers and software packages:
HP Helion Rack - Initial Configuration
- One HP DL360 server that acts as a Seed node of virtual machines (Seed VM);
- One HP DL360 server that serves as a controller for the low-level cloud (UnderCloud);
- Three HP DL360 servers that perform the functions of a controller for the top-level cloud (OverCloud);
- Two HP DL380 servers that act as Starter Swift object storage nodes;
- Three HP DL380 servers serving as VSA Cinder block storage nodes;
- Four HP DL360 servers performing the functions of compute nodes;
- Two HP FlexFabric 5700 switches;
- HP Helion OpenStack package with nine licenses;
- HP Helion Development Platform package with nine licenses;
This minimum configuration can be expanded by adding HP DL360 compute nodes and HP DL380 object and block storage nodes.
Computing nodes run custom virtual machines, the number of which, served by one physical server, directly depends on the number of processor cores and the amount of RAM. The HP ProLiant DL360 Gen9 servers, which serve as these nodes, use two 18-core processors and 256 GB of RAM, as well as a RAID 10 array of four 10k SFF drives of 1.2 TB each. In the minimum configuration, HP Helion Rack includes four compute nodes on which up to 500 virtual machines can be deployed (in practice, the maximum number of VMs depends on specific applications and their performance requirements).
The HP iLO Remote Administration Package is used to manage all rack servers, and with the help of the open APIs, HP Helion Rack can be integrated with other administrative tools.
The initial configuration of the HP Helion Rack occupies only 22U in a 42-unit rack, and in the remaining 20U it is possible to accommodate additional computing and / or storage nodes depending on the specific requirements for a private cloud (see table). You can also scale the configuration of individual servers by installing additional processors, memory and disks. In addition, HP Helion Rack scales to several racks and in the maximum configuration supports 100 nodes (in the current version).
Our previous publications:
» HP OpenStack Helion - First Meet
» HP StormRunner Load. A practical guide. Part II
» HP StormRunner Load. A practical guide. Part I
»The correct bonus for your new server
» Cost-effective HP servers for SMB and providers
» SDN: alternative or addition to traditional networks?
» New network architectures: open or closed solutions?
» Implementing MSA in a virtualized enterprise environment
» HP MSA Disk Arrays as a Basis for Data Consolidation
» Multivendor corporate network: myths and reality
HP Helion Rack is a fully configured, configured and tested platform for building a private cloud based on OpenStack, designed for the design, development and deployment of cloud applications. The main purpose of HP Helion Rack is to host applications that require a reliable server infrastructure with optimally tuned performance. It uses HP Helion OpenStack software (augmented by HP’s OpenStack Distribution Module) and HP ProLiant DL360 / DL380 Gen9 ninth-generation rack servers installed in the same rack and integrated using HP FlexFabric 5700 switches. In fact, HP Helion Rack is a “turnkey solution” »For the rapid implementation of the OpenStack cloud, eliminating the need for the customer to engage in the design of the frequent cloud, installing and configuring the OpenStack package, deploying and configuring the cloud-servicing server equipment.
Usage scenarios
HP considers the development and deployment of applications for use in the cloud (cloud native), the rapid deployment of infrastructure for developing and testing applications, and acting as a private cloud resource broker for “heavy” cloud applications as the main use cases for HP Helion Rack.
')
HP Helion Rack Architecture
The first image shows the main software and hardware components of the HP Helion Rack — the HP Helion Development Platform application platform for providing PaaS services, the HP Helion OpenStack cloud deployment platform, and the HP ProLiant Gen9 servers.
HP Helion Rack supports various scaling options depending on the target destination, so the specific hardware configuration of the servers and their number depends on the tasks of the cloud that will be deployed on them. As cloud controllers and computing nodes, the solution uses HP ProLiant DL360 Gen9 single-duplex dual-socket servers, and HP ProLiant DL380 Gen9 dual-dual two-socket servers as storage nodes, where you can install up to 12 full-size disks.
HP Helion Rack Architecture
The minimum configuration of this solution consists of the following servers and software packages:
HP Helion Rack - Initial Configuration
- One HP DL360 server that acts as a Seed node of virtual machines (Seed VM);
- One HP DL360 server that serves as a controller for the low-level cloud (UnderCloud);
- Three HP DL360 servers that perform the functions of a controller for the top-level cloud (OverCloud);
- Two HP DL380 servers that act as Starter Swift object storage nodes;
- Three HP DL380 servers serving as VSA Cinder block storage nodes;
- Four HP DL360 servers performing the functions of compute nodes;
- Two HP FlexFabric 5700 switches;
- HP Helion OpenStack package with nine licenses;
- HP Helion Development Platform package with nine licenses;
This minimum configuration can be expanded by adding HP DL360 compute nodes and HP DL380 object and block storage nodes.
The functionality of the components of HP Helion Rack
Cloud controllers
- Seed node
The Seed node is used to boot the starting virtual machine and deploy the UnderCloud cloud. The HP ProLiant DL360 Gen9 server, which acts as a Seed node, serves only DHCP and network boot services for UnderCloud and stores scripts to back up and restore other controllers, so it only needs one six-core Intel Xeon E5 processor, 32 GB of RAM, and four hard drives 3 , 5-inches, combined in a RAID 1 array (i.e. using mirroring). - UnderCloud node
Cloud UnderCloud provides deployment, management and maintenance of the cloud. The UnderCloud node must be powerful enough to handle a large amount of I / O for centralized event recording, monitoring programs, and other functions. The HP ProLiant DL360 Gen9 server, which functions as the UnderCloud node, is equipped with two six-core processors, and its RAM is increased to 64 GB. Like the Seed node, it has four full-size hard drives, but these are high-speed drives of 15k RPM of 600 GB each, combined into a RAID 10 array, providing not only the acceleration of reading and writing data from the disk, but also their protection from disk failure. - OverCloud controller nodes
OverCloud controllers serve the core services of the productive OpenStack cloud, including Nova, Keystone, Glance, Cinder, Heat, Neutron, Ceilometer and Horizon, so HP ProLiant DL360 Gen9 with powerful configuration (two 12-core Intel Xeon E5, 64 GB of RAM and four 600-GB LFF 15K hard drives combined in RAID 10). To ensure the protection of the cloud from the controller hardware failures, three separate controller nodes are used with a failover cluster.
Storage nodes
- Block storage
Cinder block storage is implemented using HP VSA-based software-defined storage on three HP ProLiant DL380 Gen9 servers, which are equipped with 10 1.2 TB disks combined in a RAID 5 array and support scaling to 26 drives. To increase the performance of Cinder storage nodes, SSDs are installed into these servers and the VSA Adaptive Optimization option is used. - Object storage
Swift nodes are used by the Glance service to store images and snapshots of virtual machines, so the main thing for them is maximum capacity, not performance. Nodes are built on HP ProLiant DL380 Gen9 servers with 15 LFF disks with capacities of up to 6 TB each. Of these disks, two are used to boot the operating system (these are 15K drives of 600 GB combined in RAID 1), and the remaining 13 are slower 7.2K drives that are combined in RAID0 or work as JBOD and provide up to 78 TB of total physical capacity on the knot. For full backup, you must use at least two storage nodes. In the current version of the solution, Swift supports scaling up to 12 storage nodes.
Compute nodes
Computing nodes run custom virtual machines, the number of which, served by one physical server, directly depends on the number of processor cores and the amount of RAM. The HP ProLiant DL360 Gen9 servers, which serve as these nodes, use two 18-core processors and 256 GB of RAM, as well as a RAID 10 array of four 10k SFF drives of 1.2 TB each. In the minimum configuration, HP Helion Rack includes four compute nodes on which up to 500 virtual machines can be deployed (in practice, the maximum number of VMs depends on specific applications and their performance requirements).
- Network infrastructure
HP Helion Rack uses a third-level switch HP FlexFabric 5700-40XG-2QSFP + Switch Top-of-Rack (ToR) to provide 40-gigabit uplinks to connect the rack to the outside world (for example, a corporate network or a public cloud via the Internet). for North-South and 10 Gigabit traffic for East-West traffic. If HP Helion Rack is used in the maximum configuration (with a 42-unit rack fully filled with servers), then you can optionally use the more powerful HP FlexFabric 5930-32QSFP + switch. This switch provides greater horizontal scaling of compute nodes and improves overall throughput when using VXLAN.
Another HP 5700-48G-4XG-2QFP + switch serves the HP Helion Rack service network, which is used to manage physical servers using IPMI and HP Integrated Lights-Out (HP iLO) tools. Through this one gigabit network, Ironic, using IPMI, monitors the status of servers and performs remote on / off servers when deployed on bare metal. On all servers serving the Undercloud and Overcloud clouds, the iLO ports connect to this switch.
Management and scaling of HP Helion Rack
The HP iLO Remote Administration Package is used to manage all rack servers, and with the help of the open APIs, HP Helion Rack can be integrated with other administrative tools.
The initial configuration of the HP Helion Rack occupies only 22U in a 42-unit rack, and in the remaining 20U it is possible to accommodate additional computing and / or storage nodes depending on the specific requirements for a private cloud (see table). You can also scale the configuration of individual servers by installing additional processors, memory and disks. In addition, HP Helion Rack scales to several racks and in the maximum configuration supports 100 nodes (in the current version).
| Knots | Computational (DL360 - 1U) | Object storage (Swift, DL380 - 2U) | Storage blocks (Cinder VSA, DL380 - 2U) |
| HP Helion Rack Initial Configuration | four | 2 | 3 |
| Maximum number of nodes per rack | 24 | 12 | 13 |
» HP OpenStack Helion - First Meet
» HP StormRunner Load. A practical guide. Part II
» HP StormRunner Load. A practical guide. Part I
»The correct bonus for your new server
» Cost-effective HP servers for SMB and providers
» SDN: alternative or addition to traditional networks?
» New network architectures: open or closed solutions?
» Implementing MSA in a virtualized enterprise environment
» HP MSA Disk Arrays as a Basis for Data Consolidation
» Multivendor corporate network: myths and reality
Source: https://habr.com/ru/post/258831/
All Articles