Essential Technology: A GCS Blog

I've often called out Google and other cloud computing vendors that promise cost savings. The truth is that the SMB market makes better decisions with their IT dollars than larger businesses. The often touted savings just don't exist! Cloud computing is new-enough and its vendors proud-enough that most offerings have a very large margin priced in. Enter SoftLayer. SoftLayer delivers a range of managed hosting options as wells as CloudLayer. CloudLayer is a virtualized, highly available infrastructure priced to compete with on-premise solutions in the SMB space.

Unlike other cloud infrastructure providers, SoftLayer features technology frequently found in the SMB. They utilize Fortinet, Dell's Equallogic and Citrix's Xen Server. You'll also find SuperMicro servers and Array Networks load balancers. A powerful web interface enables rapid provisioning and easy management across SoftLayer's three data centers. Pricing for their Public Cloud option starts at only $99 per month for 1 Core + 1GB RAM + 100GB Storage. Bandwidth is included up to 2TB of outbound traffic while inbound and server-to-server is not metered.

SoftLayer data centers are SAS 70 Type II certified and feature bandwidth or peering with many of the top telecoms. They claim 20,000 servers and more than 5700 customers. In a market dominated by enterprise focused companies, it is exciting to see a cloud solution priced for the SMB. I'm planning a trip to Dallas for the tour shortly and an extensive evaluation. I'll fill in more details when available.

Moving large amounts of data offsite is a difficult thing to do. It is difficult to beat tape drives and an automobile for rapid data replication. Obviously this method has serious limitations. With the rise of Shared Storage, most replication today occurs at the Storage layer. Though application layer replication offers major advantages, it is usually pursued in addition to storage replication.

What is storage replication? Most SANs and many NAS devices offers licensed replication as a feature. GCS' two favorite devices, the HP Lefthand and Dell Equallogic, include these licenses at no cost. By placing a SAN in-office and one at a remote data center, then connecting the sites with sufficient bandwidth, the SANs will replicate data between the two at the block level. This is an efficient method. Unfortunately, the SANs cannot differentiate between legitimate data changes (ie, a saved Word doc) and temporary data changes such as SQL log files. Everything is pushed across the connection. This tends to require large amounts of bandwidth.

The HP Lefthand offers both asynchronous and synchronous replication options. With asynch replication changed data is pushed at scheduled intervals. With synchronous replication the devices attempt near real-time replication. The HP Lefthand only offers synchronous when there is less than 20ms of latency between the sites. The HP Lefthand also offers bandwidth management on the SAN to help restrict the amount of bandwidth consumed on your site-to-site connection.

A readily apparent benefit of SAN replication is virtual machine replication. If my VMs are stored on my SAN, they are replicated to my remote sites. This dramatically reduces recovery time in a variety of outages.

GCS also recommends Microsoft DFS-R for replication of file stores. Simple, reliable and efficient, this technology allows Windows NAS devices to replicate between sites and provide failover in the event of an outage.

Petri reviews the new VMWare vSphere Data Recovery feature in a recent post. Most important to note is that there is added support for VSS for Windows VMs, no tape drive support and no agent/plugin support for SQL, Exchange, etc. This highlights one of the advantages Microsoft's Hyper-V has in comparison to VMWare, Microsoft's Data Protection Manager (DPM).  DPM is a mature, reliable backup software that offers tape support, VSS support, Exchange/SQL/Sharepoint agents and DPM-to-DPM replication. Now DPM Server is not included in Microsoft's System Center Server Management Suite Enterprise or Datacenter. However all agents for DPM are included. For another $600 you get a complete backup application.

Most of our clients choose between an Entry Level or Mid-Level SAN, as described in Part 1 of this Buyer's Guide. In this post I'll identify the features that differentiate those two classes of SANs.

Clustering: Clustered SANs operate like clustered servers. If one fails the other takes over immediately. This is only available in the Mid-Level products such as HP Lefthand, Dell Equallogic, etc. Only HP's Lefthand offers true clustering out of the box. The minimum order for a Lefthand SAN is two completely separate units.

Thin Provisioning: Mid-level SANs allow  you to oversubscribe storage by allocating storage to a volume without reserving that storage. If you create a 100GB volume but only use 40GB, the remaining 60GB is free to be allocated to another volume. This feature is essential in maximizing the storage efficiency of a SAN.

Offsite Replication: Replication between SANs is the foundation for an excellent DR solution. Replicate all data and VMs to another site. Many of the mid-level SANs offer this solution but in some it is a licensed add-on. HP's Lefthand includes scheduled replication at no additional cost but real time replication and automated failover is an additional license fee.

Snapshots: Snapshotting technology is an on-array backup method that utilizes a relatively small amount of disk space. This is possible to restore entire volumes quickly without relying on external storage.

De-duplification: Long an enterprise only feature, integrated de-duplification is making its way into mid-level SANs. This can dramatically increase the efficiency of storage but can have a significant performance cost.

Last week Microsoft published the first release candidate of the new Exchange Server 2010. This is one of the final milestones in the roadmap for final release early next year. As with any new update the biggest question is what is in the upgrade that makes the expense and hassle of an upgrade worthwhile. This is especially true for users that have recently upgraded to Exchange 2007.  I’m pleased to say that while there are numerous improvements in Exchange 2010, there are several that standout and make a compelling argument for upgrade.

At the top of the list are the improvements made for high availability.  Exchange 2007 offered a number of different high availability (HA) options (SCR, CCR, and SCC).  Standby Continuous Replication offered relatively inexpensive DR functionality for Exchange, but failover was a manual process. CCR, or Cluster Continuous Replication, and SCC, or Single Copy Cluster, allowed for automatic failover within a single site but required installing only the mailbox role inside the cluster. This meant a minimum of three Exchange licenses to achieve high availability, four if you wanted redundancy for the other Exchange roles. In addition to that expense, the configuration was a very manual process that required a great deal of time.

In Exchange 2010, Microsoft has eliminated many of the prior HA configuration in favor of a new concept called database availability groups (DAG).  DAG allows for a fully redundant configuration with onsite or offsite replication. Microsoft has changed the architecture of Exchange so that the mailbox database is no longer tied to a specific server. DAG permits a mailbox database to be replicated to up 16 different servers. What’s more the configuration has been highly simplified and can be fully managed from the Exchange Management Console.  You can even configure a single server for DAG and then add a second server at a later date without reconfiguring.

In addition to DAG Microsoft has improved the consistency of mail delivery with the concept of Shadow Redundancy.  Where DAG applies to the availability of mailbox database, Shadow Redundancy handles providing redundancy for messages during the transportation process. Basically each server in the transportation process holds onto a copy of a message until it receives confirmation that the message was received by the next “hop” in the process.  If confirmation isn’t received the message can be resubmitted to the next server or an alternate server if one is available. Unfortunately, Shadow Redundancy can only be used up to the point the mail is forwarded to a server that doesn’t support this functionality. But, it is a big step forward in ensuring consistent mail delivery within the Exchange routing environment.

Aside from high availability, there are numerous other improvements built into Exchange 2010. Some highlights include optimization for handling large individual mailboxes (10GB and larger), further improvements to Outlook Web Access (now known as Outlook Lite), and new message archiving and compliance functionality out of the box.  These are valuable improvements that I’ll expand on at a later date. But, email has become a mission critical application in organization large and small these improvements to increase the uptime of Exchange are by far the most compelling.

The Xilocore offsite backup solution came recommended from another VAR out of state. When I saw them at the CompTIA Breakaway Summit I made a point to spend some time at the booth. If you're not familar with Xilocore, it serves the same purpose as an offsite SAN replication of VMs and data without needing the offsite SAN, servers, data center, firewall, bandwidth, etc. An onsite appliance is installed that makes a local backup and then replicates the data offsite. In the event of a server failure you can run the VM on the Xilcore box locally. If a more serious or prolonged outage occurs, Xilocore can stand up your VMs at their data center and provide Citrix-based access to those servers. Turnaround times are offered at 24 and 48 hour intervals.

I looked at this solution in comparison to a proposal I was working on that featured offsite SAN replication with offsite server hardware for DR. This was a Hyper-V platform with System Center Management Suite (including DPM). Xilocore had a compelling offer but this client already had collocation, an MPLS WAN and gobs of spare hardware coming after the virtualization project completes. Though not a fit for this client, I'll circle back and investigate more on upcoming opportunities. I'll be sure to post the outcome.

Reluctantly, I've become a big fan of Avaya's IP Office product. My early impressions of this product were not favorable. However it's mobility and remote connection features have developed into best-in-class features. The IP Office has always been very reliable. Finally, it is price competitive from ten - several hundred handsets. The IP Office is getting even better with the impending release of the 5.0 software upgrade. What's new in 5.0?

Failover: IP Offices can now failover between switches either geographically or within the same location. This is only supported with IP handsets.

Exchange 2007 Integration: Voicemail integration is now supported.

Web Based Telephony: Many users that previously relied on Phone Manager Lite will be able to access similar functionality over the web.

Conference Bridge: The conference bridge can be upgraded from 64 ports to 128 ports.

Small Community Network: IP Offices can now be installed at up to 32 individual locations and networked for 3 or 4 digit dialing between locations.

SIP Endpoints: Though I'm sure this feature has been heavily requested its impact on the market will be negligible. Additional licenses are required per SIP handset. Avaya's 16xx handsets rival SIP handsets in price, do not require an additional license and provide more features.

Licensing: Avaya has simplified the licensing by bundling user and phone switch licenses by user type. There are now three types of switch licenses. Essential uses the Embedded voicemail, which now includes Dial by Name directory. Preferred adds Voicemail Pro functionality. Advanced bundles call center reporting, recorded call management, conditional call routing and more. In addition to the switch licenses there are six types of user licenses - Mobile Worker, Tele Worker, Power User, Receptionist, Customer Service Agent and Customer Service Supervisor. Mobile Worker is essentially for remote extensions without any telephony or handset use. Tele Worker includes all licenses for remote IP phone or telecommuter access. Power User bundles Mobile and Tele Worker licenses. Receptionist adds the soft console application.

Support: IP Office 5 is only supported on the 500, 412 and 406V2 switches.