What is new with OnMetal Cloud Servers
What is OnMetal?
OnMetal servers are single-tenant, bare metal servers provisioned via the same OpenStack API as our cloud. They can be spun up or down as quickly as VMs to offer the agility of multi-tenant environments with the performance of single-tenant hardware. OnMetal servers are engineered as 100 percent solid-state with external cooling. This provides increased mean time between failures (MTBF). They are very large, so you will need fewer of them.
OnMetal is designed for large or quickly-growing Internet businesses thinking about moving from colo to cloud, or vice versa:
- You pay by the minute, just like with VMs. But OnMetal servers are much simpler and more powerful than VMs.
- They are built on OpenStack software and Open Compute hardware. The API for OnMetal is the OpenStack Nova API. It's familiar to users of any OpenStack public cloud. Users don't have to worry about vendor lock-in.
- OnMetal servers are customized for specific workloads.
- There is no hypervisor, and no virtualization tax.
- There is no sharing of metal with any other user.
OnMetal combines the simplicity of consistent performance and the economy of colocation with the elasticity of the cloud. Running your high-traffic production environment on consistently performing bare metal machines means less over-engineering, more simplicity and - ultimately - lower costs. Because OnMetal is a part of the Rackspace Managed Cloud portfolio, you won't spend as much managing your servers.
The chassis is all solid-state. We removed cooling fans from the boxes and do not using any spinning media. This reduces heat and vibration, and helps increase MTBF. In order to deliver the economy of colocation that customers require, we optimized the configuration based on specific workload requirements like "database transactions per second per dollar" or "total RAM per dollar per hour." This led to the following configurations:
|10 Cores, 32 GB RAM, no disk, 10 gigabit||Latest Xeons optimized for CPU per dollar, perfect for handling web requests and doing background processing. Optimized for high CPU activity like network requests, application logic, web servers, and load balancers.|
|20 cores, 128 GB RAM, 3.2 TB of PCIe Flash, 10 gigabit||Optimized for the best performance per dollar for the most common open source databases. Optimized for high I/O activity like NoSQL and SQL databases|
|Memory||12 cores, 512 GB RAM, no disk, 10 gigabit||Optimized for the best RAM per dollar ratio or memory-heavy workloads. Especially useful for high RAM activity like in-memory SQL configurations, caching, and searching indexes.|
The fast 10-gigabit network is used for all instance types, because network performance is becoming increasingly important. All OnMetal server flavors have a 32 GB system disk. The OnMetal High I/O flavor also has two 1.6 TB PCIe flash drives, which are mounted directly into /dev as regular disks. Rackspace recommends using Linux built-in software RAID. OnMetal Compute and Memory flavors only use the 32 GB system disk and are considered “diskless,” with no data disks. Currently, Cloud Block Storage cannot be attached to any OnMetal server instance. For information on configuring the flash drives as data drives, see Configure flash drives in High I/O instances as Data drives.
API-driven instant provisioning of OnMetal Cloud Servers brings the agility and elasticity of the cloud without the problems of multi-tenant environments. All OnMetal flavors include dual 10-gigabit Ethernet (10 GigE) 10G NICs in a high-availability bonded configuration, and use VLAN tagging to access ServiceNet (for traffic within a Rackspace data center) and PublicNet (the Internet).
OnMetal servers are not offered as single-tenant VMs for the following reasons:
- Several customers expressed concerns with virtualization tax, which becomes more important as the number of servers grows. While hypervisors continue to get better, we routinely meet customers who feel the impact of this tax.
- We see a technology trend in software that renders virtualization less useful. In a single-tenant environment like OnMetal, the application management and isolation that virtualization requires is not necessary.
- The progress in operating systems has delivered a native capability to isolate applications by using containers. Companies like Docker and CoreOS provide tools to run fully isolated applications without relying on virtualization, and we see this as an emerging trend to run at scale: containers on bare metal.
Depending on your application needs, you have the option to mix and match multi-tenant VMs with bare metal instances. OnMetal servers must be created using SSH key pairs. You should neither use nor modify the administrator password generated by the server build operation. For information on generating SSH keys, see Manage SSH Key Pairs for Cloud Servers with python-novaclient.
After an OnMetal server is built, there is a slight delay in network availability. Even after the status reads ACTIVE, a few additional minutes (3 to 4) are needed for the network configuration to complete. After the new OnMetal server pings successfully, you can begin to use it.
You cannot attach a private network to an OnMetal server. Future generations of OnMetal servers will have this capability. Until then, use ServiceNet for internal traffic, and remember to secure your OnMetal server because ServiceNet is open to other Rackspace customers.
With no hypervisor, there is also not a web terminal or console as on other Rackspace cloud servers. This limits the optons available if you lose access to your OnMetal Cloud Server. If access is lost, you must reprovision your OnMetal Cloud Server. Use caution with your login information and network configuration.
The following actions and features are not possible on OnMetal servers:
- Server rebuild
- Server resize
- Rescue mode
- Server image creation
© 2011-2013 Rackspace US, Inc.
Except where otherwise noted, content on this site is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License
See license specifics and DISCLAIMER