As the enterprise-class cloud, we at HP Cloud Services take pride in holding ourselves to a higher standard. We of course have SLAs that we view as industry-leading, but that’s just part of the story.
I’d like to briefly dispel some myths about what an SLA is, and then as the focus of this post I will talk about: what HP Cloud Services is doing to actually deliver on and surpass its SLA and its enterprise-class commitment to its customers.
Contractually, an SLA is a document that provides legal remedies to a customer in the event of a service failure. That said, no customer I’ve ever talked with has asked about our SLA in hopes of getting large amounts of money rebated via an SLA. And to that point, if we’re ever in a situation where we’re rebating large amounts of money to our customers, then SLA or no SLA, we have failed our customers.
As such, it’s worth noting that while some service providers may choose to offer a “100% SLA,” all that means is they’ll give the customer some money back if there’s downtime—it doesn’t actually mean a customer will get “100% everything,” and in fact history at most cloud service providers would show that even the best intentioned provider will have a hiccup occasionally. More important to our customers than the details of calculating a legal remedy is seeing that:
1) HP has financially aligned itself with its customers (via an SLA) – so if our customers hurt, we do too – and that
2) HP has taken the necessary steps to actually deliver on our enterprise-class commitment to our customers.
So what is HP doing to deliver truly enterprise-class service along the dimensions of: data durability, availability, reliability, security, and customer support? At a high level, here are some of the things we’re doing:
For HP Cloud Object Storage, to back up our industry-leading 99.95% or higher monthly SLA, we create three replicas of every file uploaded. Each replica is stored in a physically separated Availability Zone (AZ), and each AZ has dual power grid, dual Internet provider access, and in general is designed to be a fully isolated fault domain. As a result, you could have even an entire AZ fail and not have any downtime.
Our service, based on OpenStack Swift & enhanced on the back end via HP innovation and best practices, is self-healing and checks the files for integrity continuously throughout the day. Even if an entire AZ hypothetically were to go down, the system would immediately fail over and create a new third copy in one of the remaining two AZs, until the third AZ came back online. On a related note, we continuously monitor the health of our system, both hardware and software, with the intent of detecting any potential problems before they have a chance to impact our customers.
Despite massive redundancy at the software layer, we further enhance the reliability and performance of our service by running on high-end HP hardware.
For HP Cloud CDN, we leverage the massively distributed & redundant worldwide network of the Akamai Intelligent Platform™ to avoid single points of failure. As such, even if all HP datacenters go down, the service is architected so that content already in the CDN will continue to be cached without interruption.
Like with storage, the HP Cloud Compute service runs OpenStack® Nova on high quality, enterprise-class HP hardware, to deliver high levels of performance and reliability. On top of this high-end hardware, we have built software-layer redundancies. For example, we use a wide variety of proprietary mechanisms & configurations (in our database, message queue, hard drive configurations, etc.) to make the system highly available through removing single points of failure. We continuously monitor the health of our system, both hardware and software, with the intent of detecting any potential problem before it has a chance to impact our customers. Finally, we leverage the same physically separated Availability Zones described above in storage.
Datacenter & Networking
Our datacenters are hosted in world-class facilities that offer redundant power and cooling systems, and high levels of physical security. In each region our infrastructure is divided into physically separate Availability Zones designed to operate independently of each other, each with their own fault tolerant networks.
In addition to the physical security of our datacenters, our infrastructure is monitored and protected by HP Enterprise Security’s suite of products and expertise, which includes products such as TippingPoint™ and Fortify™, as well as DDoS prevention technologies. Additionally, HP has long been (and continues to be) an industry leader in setting and adhering to global privacy laws. Finally, HP is a major contributor to OpenStack security initiatives, including responsible disclosure and proactive analysis of vulnerabilities, and we have expanded the capabilities of the OpenStack Keystone Identity Service in several areas including encrypted tokens, stronger role enforcement and optional multi-factor authentication.
In addition to our self-service Knowledge Base, our customer support team is available to talk directly to you 24/7, about any issue, technical or otherwise. They are highly trained, and they have the authority to escalate directly to our engineering directors as needed to ensure prompt resolution to any issue you may have. Additionally we are in the process of adding further enhanced, premium levels of support for our most demanding customers.
This is by no means a complete list, but rather just some of the things we are doing to ensure HP Cloud Services is the cloud of choice for those customers that value enterprise-class: data protection, data availability, system-wide reliability and security, and customer support.
As always, this is just the beginning…stay tuned.