About Cloud Clustering

The cloud economy is so compelling, today, more and more data center managers are starting to assess which additional applications are more valuable to cloud services, whether it is a private cloud that deploys its own enterprise, or whether it is a mix or a public cloud service that uses Amazon Network services or other service providers. At the same time, there is growing evidence that proven enterprise-class features are being looked at, as they provide reliable services to comply with SLA agreements.

As data center managers become accustomed to using cloud services more and more, they don't want to give it up.

Enterprise Storage

For traditional enterprise storage system, clustering storage is one of the main characteristics. It can be said to be a standard enterprise approach for achieving high availability in critical enterprise applications. Clustering runs the same application through multiple server clusters, causing any failure of any server to cause downtime (other servers will replace the failed server task). This is necessary for a database, and since many enterprise calculations are run on a database, most enterprise applications cannot migrate to cloud services without it.

So far, using cloud clustering requires enterprise IT teams to specifically overwrite legacy applications for cloud deployments because the storage system constrains the database for one or two of the following reasons:

(1) storage is too slow, requiring applications to be decomposed into parallel processes, each running on a low speed storage, but overall to ensure sufficient performance.

(2) storage lacks certain functions, such as volume sharing, NFS, or CIFS, which are common in traditional applications.

Of course, few IT departments have time to do this additional work, so applications need clustering to stay out of the cloud, or "hosting," where a service provider creates a private setting for customers who use a dedicated device. This cost is very expensive and rigid and takes a long time to revise and promise for many years.

Software-defined Storage

Software-defined storage is rapidly changing all of these conditions. Select a Third-party solution to create a cluster in the cloud, unaffected by network limitations such as lack of multiple IP groups and iSCSI persistence, and the impact of a single volume to multiple cloud servers.

Depending on the extent of these limitations, basic file sharing work together, such as CAD, CAM, and other types of shared workloads will be impossible. Clustering is a standard for most legacy applications (SQL, Exchange, Oracle). Migrating these clustered applications to the cloud leaving clustering is not an option.

The new software-defined approach can also intelligently and quickly share storage hardware among multiple customers without affecting any privacy or performance issues.

To support clustering, a cloud-based storage approach requires a list of features:

L application or operating system cluster support (for example, a red Hat cluster failover or a failed cluster transfer for Windows).

L capacity sharing allows the same volume volume to be installed on all servers in the cluster, rather than allowing only one volume to match a single server attachment, as in the case of a common cloud storage solution.

L Support SCSI persistent retention to prevent the server from modifying the same data at the same time. Essentially, each server can temporarily lock other servers from one data area to keep the first server working, a feature that is common to storage systems deployed in-house, but rarely in emerging cloud computing.

L multi-IP Group or 2-tier communication supports clusters between servers to determine the normal operation of each other.

In addition, storage needs to increase performance as the number of servers in the cluster increases, and each server generates more I/O, which in itself can be a challenge because these applications often start with the nature of IOPS.

Deloitte Consulting's infrastructure and operations hosting business has been using dimension Data's server, storage-as-service (STAAS) solution and its accompanying multi-IP capabilities since 2013. The Dimension data company supports Zadara storage for native SQL clusters. In addition to using Zadara to deliver more than terabytes of storage capacity to a wide range of customers, Deloitte Management Analytics SaaS Products also employ Zadara to provide cloud-environment information Insights, and Deloitte hosted client hosts to help organizations run their IT infrastructures more efficiently.

Some customers have 4 9 ' s or 5 9 ' s levels of SLAs, some of which are more demanding, including retailing, inventory management and patient management, and some customers say they have only a small space for cloud-induced systems. Often, Deloitte will take a hybrid approach by setting up the network layer and application layer in cloud computing and keeping the physical servers in the field connected to the cluster storage resources. But doing so means considerable maintenance time and storage costs. Deloitte has to buy storage, usually more than it needs, and when this happens, make sure to physically connect new resources to the existing network scale.

"By clustering in the cloud, enabling them to take full advantage of physical resources, they are able to get the elements agile while satisfying customer needs," said Michael Heiss, a Deloitte consulting firm's infrastructure and operations master server expert. Otherwise, the purchase and installation of physical resources in the field can not meet the growth of customer demand.

With technologies such as cluster clustering such as clouds, the previously impossible computing power of the enterprise is now possible. All IT resources will become more flexible, high-performance, and more economical, even a considerable amount of high availability has become a reality. Data center managers should be aware that they do not have to compromise between the enterprise's internal deployment and the need to keep management running an optimal operational requirement.