Single Blog

We attended this dutiful event in the beautiful Sheraton Golf2 Hotel in Rome. Below is the agenda:

• 9.30 Registration of participants

• 10.00 Welcome and opening of the meeting
  Luigi Marcocchia, Softway

• 10.10 How to harness the power of virtualization to transform Data Centers into simplified cloud computing infrastructures, capable of providing advanced, reliable, flexible and secure services Today, companies are increasingly dependent on IT services and the opportunities offered by virtualization technologies declined in the Data Center sector offer all companies (from large companies to SMEs) clear advantages from various points of view. In fact, the ability to abstract the applications and data of the underlying infrastructure, to create an internal cloud infrastructure, allows not only to increase the efficiency of delivery of existing IT services, significantly reducing the costs and complexity associated with the management and maintenance of the infrastructure, but drastically reduces the implementation time of new services. This frees IT managers from the complexity of servers, storage, and network infrastructure, allowing them to focus on realizing value for the business. What are the architectural aspects of the new vSphere solution for virtualized infrastructure? What is the concrete impact on the business of companies? Sergio Cimino, VMware

• 10.50 Towards the Unified Computing System: the new architecture for the Data Center of the future The possibilities offered by virtualization technologies to combine the increased processing capacity of IT systems, with the new performance of the network and with rapid access to storage resources, have laid the foundations for defining the evolution of the Data Center of the future towards a vision of the Unified Computing System. What are the main components of this new architecture and how do they integrate with each other? How are the demands on the physical network infrastructure changing and what are the benefits of the innovative Unified Fabric, Fabric Ethernet for IP & Fibre Channel approach? How is the traditional role of the switch evolving in a virtualized environment? What are the technological innovations taking place? How can we have the same benefits in terms of ease of configuration and operation as the traditional network services available in “physical” switches, applying them to new virtual machines? How to implement a storage consolidation process and how to correctly assess its impact in terms of costs / benefits? Roberto Missana, Cisco Italy and Vice Chairman SNIA Italy

• 11.30 Coffee break

• 11.50 Storage in a virtualized architecture In order to take advantage of all the advantages related to virtualization, it is essential to start thinking no longer about individual implementations, but about an overall Virtual Data Center design. The speech will illustrate new types of solutions and new models of offer for the storage and management of data and business information including, to highlight the main ones, multi-protocol storage, virtual tape libraries, deduplication, snapshots, thin provisioning. Roberto Patano, NetApp

• 12.30 Questions and answers

• 13.10 Closing of the meeting

• 13.20 Lunch

VMWare

The speech by Sergio Cimino of VMware focused on the news of the new vSphere 4 platform (Cloud OS) and on the performance documented by impartial benchmarks. Of particular note are these new features.

vCompute

The new ESX 4.0 hypervisor enables VMs up to 8 VCPUs, 256 GB RAM, 20 Gb/s of networking, and over 200,000 IOPS with a latency of less than 20 microseconds. The ESX can scale up to 64 cores and 1TB of RAM on hardware.

Considering the following scheme:

We see that there are few applications that would be able to harness the power of modern servers, so the virtualization layer, the hypervisor, will become almost essential in the immediate future. An example of a benchmark run with SPECweb2005 on an HP Proliant DL585G5 machine with 16 Cores with the ESX3.5 hypervisor gave a result of 44,000, i.e. able to support 3 billion page views per day, i.e. 3 times the ebay traffic on a single server. A new DPM (Distributed Power Management) feature gives the infrastructure a much more GreenIT look as DRS (Distributed Resource Scheduling) can turn off underused physical servers on occasion and then turn them back on when necessary. The possibility of being able to change the number of VCPUs and VRAMs of each individual VM is also fundamental, obviously if the virtual operating system allows it, otherwise the blue/purple screen are a must.

vStorage

As far as the storage situation is concerned, there is a considerable optimization of space, it only takes up what is really used and the disk extension can be done hot.

vNetwork

On networking there are big news, thanks to the collaboration of cisco, the network becomes virtual at the datacenter level not only at the physical machine level, this avoids significant reconfigurations, or copies of configurations between all the ESX servers in the farm.

Automatisms and control

These new features are accompanied by the possibility of being controlled by means of automatisms programmed through workflows

Here is also the possibility that we have mentioned in the past of VM fault tolerance

Certainly a very useful feature in cases of strict business continuity, but it takes up twice as many resources as a single VM.

vApp & IT as a service

The possibility of being able to define applications (sets of VMs connected according to architectural logic) capable of self-describing allows you to be able to publish/migrate the vApp between Cloud infrastructure

Cisco

Nexus

Roberto Missana of Cisco focused his speech on Cisco’s new vision of Unified Computing, which we mentioned in a previous article . The rise of virtualization is increasingly leading to requests for networks with greater performance and available bandwidth, and 10G Ethernet is also starting to be seen on server motherboards. Cisco abandons the Infiniband path to follow 10GE and FCoE (Fibre Cloud over Ethernet). The Nexus family of products enables virtualization and consolidation of the IO. The nexus 1000v, an appliance for ESXs, is the distributed switch layer mentioned above

With the logic of the Unified Fabric, the servers for virtualization if previously required about 8 network cards, now two CNA (Converged Network Adapter) are enough, which allow all the protocols to circulate unchanged in ethernet frames, thus transforming the datacenters into “ethernet datacenters”.

The feature of fundamental importance for us is the possibility of gradually migrating the entire existing network without too much trauma by eliminating useless cables from time to time, converging everything in the ethernet network, with a huge reduction in TCO also considering the possibility of reallocating network specialists to other activities, or even reducing the staff, for the most drastic.

WaaS

Another fundamental topic for virtualization, since it allows an enormous centralization of the infrastructure, are the so-called network optimizers, or accelerators. Wide Area Application Services (WaaS) accelerates the performance of any application across the WAN, a boon for remote offices and desktop and application virtualization. A feature that we found extremely nice and useful is the possibility of having a Windows 2008 virtual server inside a WaaS, thus capable of being a local Domain Controller and / or print server etc for local users, thus allowing to eliminate all the remote server infrastructure.

Unified Computing

Cisco’s Datacenter Vision Is One System

• A unique system, which includes:
  Processing: Industry standard x86
  Network: Unified fabric
  Optimized virtualization
  Storage Access: Wire once for SAN, NAS, iSCSI

• Integrated Management
  Increase scalability without adding complexity
  Dynamic Resource Provisioning
  Integration with a broad ecosystem of partners

• Efficiency
  Fewer servers, switches, cards, cables
  Less power & cooling
  Increased processing efficiency by removing I/O and memory bottlenecks

The advantages are summarized in the following table:

NetApp

NetApp’s Roberto Patano emphasizes the efficiency and flexibility of NetApp’s storage systems, saving space, energy, redundancies, security, centralized management, and how a variety of multi-purpose storage systems can be managed as a single system. The fundamental rule of the strong offer policy of NetApp devices is that they all mount the same operating system (DataOnTap), with all the features already integrated, already active or can be activated through licenses. This makes it very easy to scale systems without having to do the slightest migration, just replace the controller with the most performing version in the worst case and nothing more.

Fundamental for NetApp is the snapshot, a term coined years ago by NetApp, all the features use this technology, this snapshot of the status brings with it speed and considerable advantages and no loss of performance.

• Deduplication: Saves up to 95% space for full backups and 25% to 55% for most data sets
• Snapshot Copies: NetApp Snapshot Copies do not require “copy” space, are useful for local backup purposes, and offer up to 80% savings
• FlexVol: Thin Provisioning, Typical savings between 20% and 33%.
• Resource-constrained replication: Disk-to-disk data protection saves up to 95% space
• Raid-DP: Double Parity Raid,Saves up to 46% compared to mirrored data or RAID 10.
• FlexClone: The savings correspond to the size of the original data sets minus the blocks that are later modified in the clones

Another element that makes great merit is the SnapManager for Virtual Infrastructure, a plugin that can be installed on vCenter that allows you to control the entire storage system directly from the VMware management tool. A comparison chart by Oliver Wyman shows that NetApp is much more efficient than its big competitors:

Finally, we show a Gartner chart on NetApp’s positioning in the market