Around The Block Blog
by John Dias, Storage Architect, Dell Compellent — November 22, 2010
The future just got a little more "cloudy" for Compellent with the release of some major updates in Storage Center 5.4 including some significant hardware updates—there's a lot to take in so let's get started.
Most notable among the hardware updates with the 5.4 release is the next generation controller in Compellent's lineup, the Series 40. The updated platform brings greater IO card capacity (six PCI-E 2.0 x8 slots) and 25% better performance. This platform will allow customers to take advantage of some of the new IO options (which will also be supported on the Series 30) and scale single systems to even larger capacities. The front end IO options now include FCoE for converged networking support. On the back end 6Gb/s SAS IO cards and enclosures are now shipping, including a 2.5" 24-bay 2U enclosure with three new drive options (146GB 15K; 450GB and 600GB 10K).
On the software side there's a game-changing capability that will resonate with enterprise customers. For years, customers have enjoyed Compellent's Data Progression feature, which provides an automated tiering capability to balance capacity and performance in a given system. Within a system the data is managed in a couple of ways—data protection levels (RAID 10 for writes and distributed parity for read only data) and disk tiering (data management based on disk performance characteristics). This allows customers to grow storage as needed (either by adding performance or capacity tiers) delivering an efficient scale up storage growth model.
However, this "two dimensional" data management scheme had previously been isolated to single systems. Compellent gave customers the ability to manage multiple systems under one console but the Data Progression features described above were limited to a single system (for the most part, replicated volumes can have a different storage profile—but you get my meaning).
Traditionally, as storage environments grow, so do the number of systems. These silos create a new level of inefficiency and management complexity that potentially cancels out any benefits gained from tiering. Moving data between siloed storage is time consuming and complex and as the infrastructure expands to meet demand, pockets of IO and throughput and capacity disparity begin to appear.
Compellent has introduced a "third dimension" data management capability through Live Volume which allows for rules based automated movement of data between storage systems non-disruptively. Live Volume plays a big foundational role in Compellent's delivery of scale out storage architecture to the enterprise and the cloud.
Using Enterprise Manager 5.4 the storage administrator can designate replicated volumes as Live Volumes. This allows the target volume in the replication scheme to become a mountable real time read only volume. Writes continue to be ingested at the source system and replicated to the target.
Should the workload shift (IO or throughput) to the target, the Compellent Storage Centers will swap roles—the target site will then become the source. Workload triggers are customizable as is the setting for allowing write ownership to transition again. Live Volume also allows for manual swapping as well. Additionally, each copy of a Live Volume enabled volume may have its own replay (snapshot) schedule.
When combined with a virtual server infrastructure the possibilities are pretty exciting. For example, a fully virtualized (compute, network and storage) infrastructure can be distributed among different campus data centers or within different power/cooling zones in a large enterprise data center. As workloads on the virtualized compute environment shift the storage workload will follow. I know from experience that data center power maintenance and generator tests being a huge pain to schedule, coordinate and execute. With this kind of a compute/storage grid available maintenance windows become practically non-existent turning your internal cloud into a 24-by-forever IT engine.
Live Volume also provides higher availability for servers using multipath SAN connections. Not only can a server have two or more connections to the SAN, with Live Volume, the targets can be distributed between two systems hosting Live Volume copies. Connections to the secondary Live Volume system could be set up as failover connections. Personally, I'm hopeful that MPIO drivers in the future will become Live Volume aware and distribute read IO between Live Volume pairs—but that's me talking, not a Compellent offering.
This new three dimensional data management model extends Compellent's value into the enterprise and cloud IT offerings where flexibility, reliability and scalability are required, not just desired, features of a modern storage system. Compellent has been the leader in scale up storage and now with Live Volume they're delivering on scale out. You're going to see Compellent continue to build on Live Volume to enable customers to scale beyond.
P.S. If you're wondering about Compellent's potential delivery of data management for the "fourth dimension" then consider that Compellent is a charter member of the
Active Archive Alliance
.
Editor’s note: This blog post is also appearing on John’s personal blog, StorageGumbo.com.
by Greg Scott, Strategic Initiatives Manager at Intel Corporation — November 17, 2010
Interest in cloud computing is growing rapidly as organizations look for new ways to reduce costs and increase the flexibility of IT. For businesses building a private, internal cloud or planning to offer external cloud services, implementing the right storage solution is essential for maximizing that flexibility. Companies need storage solutions that support an on-demand model, providing the right resources at the moment they are needed without excessive costs.
As the Intel Architecture Group and Intel IT showed in the “Cloud on Wheels” demo at Storage and Networking World in October, Compellent Fluid Data Storage provides a strong platform for cloud computing environments. The Compellent solution, powered by Intel® Xeon® processors, offers a virtualized storage area network (SAN) with capabilities that deliver the cost-effective flexibility users expect from cloud environments.
Rapidly scaling capacity with thin provisioning
Like server virtualization, storage virtualization can play a key role in cloud computing. By pooling storage resources, a virtualized solution such as the Compellent SAN can help ensure that applications have the capacity and performance they need, when they need it.
Thin provisioning capabilities, which are integrated into the Compellent SAN, can help optimize capacity utilization within that virtualized environment. With thin provisioning, administrators no longer have to anticipate and pre-allocate storage capacity for each volume. Instead, they can create virtual volumes of any size—applications consume capacity only when data is written. Especially useful in cloud infrastructures, thin provisioning enables administrators to rapidly expand or shrink volumes as usage levels change without having to purchase additional arrays.
Sustaining performance and controlling costs with automated tiered storage
Building a multi-tiered cloud storage environment can help optimize storage performance while controlling costs. With a multi-tiered environment, organizations can take advantage of multiple drive types with varying performance levels. Frequently accessed data stays on high-performance drives, such as Intel® Solid-State Drives, while archival data is moved to more cost-effective drives, such as SATA drives.
With the Compellent SAN, data is written first to high-performance drives by default to ensure the best performance. Automated tiered storage capabilities then automatically migrate block-level data from one tier to another—or one RAID level to another within a tier—based on administrator-set policies. The migration process runs in the background without affecting data availability or application performance. Many organizations can move as much as 80 percent of data to cost-effective drives, helping them cut costs while delivering high performance to the “hot data” or critical data. The benefit, is that the tiering allows highly accessed data to be placed on the performance tier dynamically, without having to configure or provision the data specifically.
Intel Xeon processors help the Compellent platform deliver the intelligence for automated tiered storage. By combining outstanding compute performance with large I/O and memory bandwidth, Intel Xeon processors enable the Compellent controller to continuously collect information about data usage and perform automated data migration without a significant impact on storage performance.
Building a better cloud
Intel is committed to helping organizations capitalize on the vast potential for cloud computing. In addition to providing key technologies for cloud solutions, Intel has launched the Intel® Cloud Builder program to streamline the construction of cloud environments by sharing reference architectures and best practices. As Intel, Compellent, and other companies are already showing, optimizing cloud storage can go a long way toward maximizing the flexibility and cost benefits of cloud computing.
For more information
Editor's note: Greg Scott is a strategic initiatives manager at Intel Corporation who is contributing to the Intel storage strategy for cloud computing. Intel is one of Compellent's technology partners. To read more about the "Cloud on Wheels" demo, check out this press release and blog post.
by Shane Burton, Microsoft Product Specialist, Dell Compellent — November 09, 2010
On-premise private clouds are becoming more popular among the enterprises we work with. These enterprises are virtualizing their storage and server infrastructures to increase efficiency and flexibility while reducing costs. To quickly and effectively manage virtualized IT resources, leading technology vendors, including Compellent, are making it easy by integrating with the Microsoft System Center Virtual Machine Manager (SCVMM) Self-Service Portal 2.0 Toolkit, a free partner-extensible cloud solution offered by Microsoft.
How does the Self-Service Portal 2.0 work?
The Self-Service Portal pools data center infrastructure resources – from networking to servers to storage – and makes them readily available to individual business units. Using Microsoft PowerShell scripts with Hyper-V and storage systems that integrate with these technologies, such as Compellent Storage Center, organizations can automate the deployment of their virtualized server and storage resources.
It’s all about efficiency and speed. From a data center management standpoint, what the business unit administrator wants to be able to do is quickly provision virtual machines (VM) without a lot of management overhead. That’s where the Self-Service Portal 2.0 comes into play. It provides a web-based interface where administrators can create whatever virtual machine they want, using pre-determined configuration options and pre-loaded VM templates.
Once a request from the business unit administrator is submitted, the system automatically notifies the IT administrator, who simply needs to validate the request and initiate the provisioning process. The Self-Service Portal takes it from there, utilizing built-in automation from Microsoft, Compellent and other DDA partners to rapidly deploy the requested IT services. Advanced PowerShell scripts reduce the manual steps associated with provisioning Windows Server/ Hyper-V VM resources, expediting deployment, ensuring accuracy and allowing administrators to focus on other important projects.
What Compellent brings to the table
Compellent comes to the table with a couple of things. Number one, its advanced storage technology is robust enough to do some pretty complex things from the storage layer at the same time reducing the complexity with operational efficiency for IT administrator. Number two, we’ve incorporated the OS-level integration with Windows Server/Hyper-V into the PowerShell piece. So that’s a huge enabler for us. Not only can we talk to our storage, but we can also talk to the OS itself without requiring special drivers to connect to the server host.
Compellent has created the Compellent Solution Pack for SCVMM SSP 2.0 that essentially incorporates two PowerShell scripts for the Self-Service Portal – one that creates one or more VMs using Compellent Instant Replays, and one that deletes VMs with automatic cleanup. We plan to offer this solution pack later this quarter.
The fluid nature of PowerShell automation
Traditionally, when data center administrators provision Hyper-V VMs using a template in SCVMM, the system created the VMs on the host itself using VHD copies via the LAN. The VHD is already sys prepped, providing a ready-to-deploy template complete with OS and applications. However, since each VM consumes just as much disk capacity as the base OS image – e.g., 10 GB – the provisioning process is time consuming and wastes disk space. Creating just one VM with this approach takes 5 to 20 minutes and, in this case, utilizes 10 GB or more of capacity.
With the Compellent “create VM” PowerShell script developed for the Self Service Portal, administrators can deploy the same Hyper-V VM in 20 to 30 seconds once the VHD has been sys prepped. Instead of creating full copies of the VHD via LAN, the script utilizes a gold image captured on the SAN as a snapshot. Each VM created shares the same gold image, and the only capacity consumed beyond that base image is the negligible space required for the unique characteristics of each VM.
For example, a small organization with four department that each need 25 physical boxes to do their work would traditionally be required to fill out a procurement order, put the OS on the hardware and ensure that there is enough energy to power 100 machines in the data center. The entire process would take about a week. With Compellent and the Self-Service Portal, it takes this organization 20-30 seconds to deploy a virtual machine. When deploying 100 machines required by the business, 12GB would be taken by the OS on each machine, meaning that an additional 1.2TB would be taken up just by the OS across the 100 machines. The same scenario on the Compellent would add up to 112 GB of space for the OS.
If changing business needs call for one or more VMs to be removed, the data center administrator can utilize the “delete VM” script. This script not only deletes the VM, but also performs a complete cleanup process – from removing the volume associated with the VM to unmounting the VHD and rescanning to ensure it has been removed.
From the beginning to the end – from the application and the OS to the storage layer, this process is completely fluid and abstracted from the general user. A few clicks on the self-service portal page and the business unit admin is all done. That is what Fluid Data is all about.
To learn more about the Self-Service Portal, please read the Microsoft blog.