Posts Tagged ‘cloud’

Public, Private and Hybrid Clouds

February 28, 2012

So, what is the difference between Public, Private and Hybrid Clouds?

The terms get mixed up frequentlyImage and many people I speak with still are confusing and using them inaccurately.

In its current form, cloud computing has been around for a few years now, yet many are still confused by the terminology used to describe the various types of clouds. This has not been helped by private software and hardware vendors inventing terms to attempt to appeal to corporate IT departments everywhere. As they saw cloud computing taking off, they attempted to exploit fears as to its security by offering solutions that really don’t have anything to do with the cloud except for a made-up name. Here, we’ll attempt to cut through the noise and give a basic primer as to the types of clouds and their potential uses.

Public Cloud

When most people speak of cloud computing, this is what they mean. A public cloud is formed when a provider, someone like Amazon, Google or a smaller company, makes computing resources, such as processing power, memory or storage, publicly available over the Internet. In a public cloud environment, the user pays no bandwidth or hardware costs and setup is usually quick and easy. Although the user does not pay these costs, they usually do pay for the resources they use. Think of it like paying for only the amount of electricity you use or the amount of minutes you use per month on your cell phone. Some providers also charge a subscription fee as well. If you need more resources, the cloud can instantly provide them. There’s no need to install additional hardware or software.

Public clouds typically run on open-source software to facilitate the movement of such vast amounts of data. However, as an increasing number of software companies, such as Microsoft and Oracle, have attempted to enter the cloud computing arena, they have started to provide cloud infrastructures that utilize proprietary software. This has been a sticking point with cloud computing purists, who don’t regard public clouds running proprietary software as truly public.

From its inception, the major vulnerability of the public cloud has been security. Once your data enters the cloud, it can circulate through dozens, hundreds or even thousands of systems. This is truly frightening for anyone running applications that involve highly secure data such as financial information or corporate intelligence. And this, more than anything else, is what brought on the other types of cloud computing that are in use today.

The Private Cloud

The term “private cloud” started when hardware and software companies were looking for ways to jump on the cloud computing bandwagon while maintaining usage of their existing systems. Knowing that IT departments were nervous about using public clouds due to security reasons, these companies hit upon the term “private cloud” as a buzzword to describe a computing infrastructure privately held by a corporation that had capabilities similar to a cloud but was completely internal and thus more secure.

Cloud computing enthusiasts point out that private clouds consist of privately held devices, such as storage arrays and servers, which needed to be built and configured by the organization. This mitigates most of the benefits of cloud computing. However, companies can use virtualization to simulate some of the resource allocation features of the cloud and thus save on costs. In general, a private cloud is not really a cloud at all but simply a farm of internal resources that can be used only by the organization in which they are installed.

The Hybrid Cloud

The hybrid cloud is the happy medium of cloud computing. If an organization has varying needs regarding computational resources and also has both sensitive and non-sensitive applications, it can use a hybrid cloud to get the best of both worlds. In most cases, the database servers, which generally contain sensitive information, are kept on a private cloud, and a public cloud is used for everything else. This solves the security problems of public clouds and lets an organization take advantage of all that public cloud has to offer when it comes to general computing resources.

As cloud computing is refined, the public cloud will become increasingly secure, allowing corporations to gradually transition their services to the more cost-effective alternative. Until then, however, hybrid clouds are likely to be the most common form of corporate cloud computing.

Effective Cloud disaster recovery

December 12, 2011

As the cloud gains steam and expands, disaster recovery has become more effective. Creative ideas involving replication have evolved as it pertains to recovery solutions for disasters.  Not always inexpensive, the best solutions provide almost instant recovery times.

Continuous data protection (CDP) using virtual machines for data replication in the cloud is probably the most reliable solution to date.

(For the purposes of this discussion, bandwidth is plentiful and security procedures have been established.)

There are two very good solutions for CDP disaster recovery on virtual machines:

1. Pure cloud  

This solution is straight forward. If you’re running your applications purely on the cloud, with nothing local, then the managed service provider (MSP) can be responsible for disaster recovery. If the primary cloud site fails, a secondary cloud site will take over with a “flick of the switch” of duplicate data and virtual machines running the applications. After recent outages at Amazon’s AWS, MSPs are ensuring they have more reliable disaster recovery solutions.

2. Replication to virtual machines from local systems

This solution works well for companies that want their data on the premises, as well as in the cloud. There are a few steps:

a.   Service provider installs an on-premise device that replicates all local data.

b.   On-premise system replicates with virtual machines in the cloud.

c.   In the event of an on-premise disaster, the “switch is flipped,” and the virtual machines in the cloud take over.

Cloud software capable of replication include CommVault Continuous Data Replicator, EMC Atmos and the Hitachi Content Platform (HCP). IBM, AppAssure, Iron Mountain, and Simply Continuous also provide these disaster recovery cloud services.

Remember, these solutions must be clearly identified in any service-level agreement (SLA) with the MSP. A key element will be the recovery time objectives (RTO). How long can the system be down before the business is impacted?  There are some key areas you want to make sure are covered, and that you fully understand, to make sure they meet your business needs and that no surprises come up should disaster strike.

1. Read the Service Level Agreement (SLA)

Read and understand the SLA being offered by your cloud service provider. Understand what constitutes a disaster, ask questions and walk through scenarios to be sure you fully understand what constitutes a disaster and the specifics around a DR event. Who declares a disaster, what processes and technologies are in place to minimize the impact to customers, and how long will it take to restore service. If a few hours of downtime seems like an eternity, and if your business cannot survive it, a cloud service may not be right for you.

2. Recovery Point Objective

The word disaster implies that bad things have happened, and when it comes to an IT service, that usually implies data loss. Make sure you understand the recovery point objective of the service so you know just how much data loss is possible in the event disaster strikes.

3. Recovery Time Objective

Communicate to all stakeholders the time that has been agreed to in your Service Level agreement for recovering from the disaster.  Make sure that all stakeholders have agreed to the recovery time before signing up with the cloud service.