Meraki and the The Cisco Cloud Networking Group

I think this is the most interesting part of the acquisition;

"Cisco’s strategy is to take Meraki’s cloud platform and business model and scale this within Cisco as our new Cloud Networking Group, led by Sanjit, John, and Hans."

I wonder if any existing Cisco groups will be moved into the Cloud Networking Group.

Cisco Announces Intent to Acquire Meraki

Hilton Romanski | November 18, 2012 at 5:34 pm PST

Cisco is dedicated to innovation as the path to growth as well as the key to sustaining our market leadership position. Our build, buy, partner strategy has always been driven by customer need and on capturing market transitions.

Today, we are excited to announce an important acquisition that addresses the rapidly occurring shift to cloud networking as a key part of Cisco’s overall strategy. San Francisco-based Meraki, a leader in cloud networking, offers customers on-premise networking solutions that are centrally managed from the cloud.

When compared to other opportunities, Meraki built a unique cloud-based business from the ground up that addresses the broader networking shift towards cloud, not just within wireless. Meraki created a massively scalable architecture that offers easy to deploy, secure, and manage networks. They didn’t obsess about the number of features, but instead focused on those that could be simplified or removed entirely.  Customers liked what they saw, and today they are supporting 20,000 customers and hundreds of thousands of network devices on their cloud platform. This has resulted in a business that is growing exponentially with great margins.

Talent is one of the most important components of every Cisco acquisition. Meraki’s co-founders, Sanjit, John and Hans, are true visionaries and leaders. The founders began with the technology, and then experimented with different markets -- pivoting from a research project at MIT to a municipal Wi-Fi company to a leading cloud networking company focused on the midmarket. Along the way, they recruited experts and created a culture in San Francisco that attracted great talent. They have focused this team around a business model that combines a rapid development methodology tightly linked to a go to market engine.

During the course of our interactions, we quickly realized that Cisco and Meraki’s shared a vision of accelerating the adoption of cloud within networking as a means to simplify operations and enable new network applications. Sequoia Capital, an early investor in Cisco, also recognized the strength of the people at Meraki, and it’s great to see the technology ecosystem come full circle.

The Meraki acquisition is another example of Cisco’s focus on accelerating our adoption of software based business models. In fact, Cisco’s last seven acquisitions (Cloupia, vCider, ThinkSmart, Virtuata, Truviso, ClearAccess and NDS) have all been software companies. Cisco’s strategy is to take Meraki’s cloud platform and business model and scale this within Cisco as our new Cloud Networking Group, led by Sanjit, John, and Hans.

I am delighted to welcome the Meraki team to the Cisco family, and look forward to a prosperous and industry-transforming future together.

Home

Administering VMware Site Recovery Manager 5.0 - Book Review

I received a copy of Mike Laverick's "Administering VMware Site Recovery Manager 5.0". This is a terrific book as the first book from VMware Press. Mike's has been providing terrific guides, white papers, and videos for years on his website RTFM Education.

To some the organization and presentation of this book may seem unconventional. Chapter 1 describes Site Recovery Manager, DR technologies, and addresses misconceptions of VMware technologies often thought of as DR technologies. Chapters 2 - 6 individually explain how to configure Dell, EMC Celerra, EMC CLARiiON, HP StorageWorks, and NetApp storage for VMware. Chapters 7 - 16 then cover the configuration and operation of VMware SRM. Chapters 1 plus one of 2 -6 make this book worthwhile to anyone installing a VMware solution with a SAN.

With my background being long in the teeth with networking and a little green in virtualization, Chapter 1 was most significant to me. I have been trying to understand the architectural differences and benefits of the different VMware technologies such as vMotion, High Availability Clusters, Fault Tolerance, and SRM.

---

Chapter 1, What is SRM, How Was DR Done Before SRM, and What VMware Technologies are Not DR

Chapter 1 provides an "Introduction to Site Recovery Manager". The chapter offers what is new in Site Recovery Manager 5.0 and, as Mike puts it, "what life was like before virtualization and before VMware SRM".

The original DR strategy was to have physical servers at the production and DR locations and rely on conventional backup and restore. For the next approach brought in virtualized servers, some suggested P2V technologies to synchronize physical servers with virtual servers. Either the production servers or the DR servers were virtualized. This approach requires the use of storage vendor's replication or snapshot technologies. This is needed to replicate the data files that make up the virtual machine's (VMX, VMDK, NVRAM, log, Snapshot, and/or swap files). Mike goes on to detail the other technology (changing IP addresses) and political issues (the storage group may not be the same people as the virtualization group) which need to be addressed. Again as Mike says, "It was again within this context that VMware engineers began working on the first release of SRM".

Next, this chapter discusses "What Is Not a DR Technology". The discussed VMware technologies provide some terrific benefits, and each have their place, but it is argued these should not be construed as DR Technologies.

As an example, some consider vMotion a DR technology. vMotion allows for virtual machines to be moved from one host to another. For vMotion to be remotely considered a DR technology, virtual servers need to be moved from one physical location to another. It is not acceptable DR design to have a DR data center within close proximity of a production data center (close enough to have your own fiber run, or where I live, to have two buildings along the same potential tornado path). To be considered a DR technology, vMotion needs to support moving virtual machines across some distance (I consider a minimum of 30 miles is necessary). Another necessary concept to understand is a vMotion is a planned event. That is, an administrator must initiate a vMotion, in a disaster scenario this is often not possible.

Finally (there is a lot in the 1st chapter) there is a discussion of the principles of storage management and replication. He does a great job of breaking through the "marketing speak" to generalize on the storage technologies most vendors support. In other words, Ford, GM, and Chrysler each offer Park, Reverse, Forward, and a radio, they may have entirely different methods of delivering these, but they all do it.

---

The Storage Vendors Chapters

Chapters 2 through 6 are dedicated to configuring specific vendors storage to work with VMware. Being somewhat new to VMware and also working in a place where I am exposed to multiple storage vendors, I really appreciated these chapters. These are great from those with limited experience configuring VMware to work with different vendors SANs. For me, these chapters were excellent. Mike provides terrific information for Dell, EMC, HP, and NetApp SAN platforms.While this doesn't cover every storage vendor, the basic principles apply to those not covered.

---

Installing, Configuring, and Customizing SRM
Chapter 7 explains installing SRM and thoroughly discusses planning and design, storage replication, and networking requirements. New VMware 5.0 features like automated failback, vSphere Replication, and bidirectional protection definitely add to the value and functionality of SRM. This chapter is very insightful for understanding the configuration of protected and recovery sites, storage replication planning and design, and configuring SRM workflow and recovery plans.

Mike walks the reader through the entire installation and configuration process with plenty of screenshots and real world examples. It is easy to follow along as he builds out a SRM solution. As the solution is built out, it covers advanced topics like customizations, scripting, and complex configurations.

The final chapter documents upgrading from SRM 4.1 to 5.0 which would be very helpful for readers still running VMware 4.1.

---

Summary
This is a terrific book from VMware Press. Mike Laverick has provided a well written and organized book. The chapters covering Dell, EMC, HP, and NetApp Storage Arrays are terrific. Administering VMware Site Recovery Manager 5.0 should be on the bookshelf of VMware and Storage admins.


Disclaimer: I received a complimentary copy of this book from VMware Press. I am not being compensated for this review. All views expressed are my own.

Radia Perlman Talk on TRILL and Spanning Tree

I found this YouTube Google Tech Talks presentation by Radia Perlman. She is often referred to as the "Mother of the Internet". She invented the spanning tree algorithm. She also invented concepts that made "link state routing" stable, scalable, and easy to manage. The protocol was adopted and renamed IS-IS. She is credited as creating the original concept of TRILL.

Her presentation is titled "Routing Without Tears; Bridging Without Danger". She discusses the creation of spanning tree, link state routing protocols and finally TRILL or Transparent Interconnection of Lots of Links. Those of of working with network infrastructure and Cloud Computing can really appreciate everything she has done.

---

Cloud Computing and Data Center Predictions for 2012

---

Solid State Drives Go Mainstream 
The terrible flooding in Thailand has caused a shortage in hard drives. Hard drive prices have increased. I expect price increases from SAN manufactures such as NetApp, EMC, HP, and others. These shortages will also increase costs for Cloud Computing providers like Rackspace, Google, and Amazon.

Credit
www.soultravelmultimedia.com 

As reported in the Wall Street Journal, LSI CEO Abhi Talwalker said that companies are turning to solid state drives to relieve some of the hard drive back log. SanDisk, Marvell, and Micron have stated they have seen or expect to see increasing demand for SSD. InfoWorld reported on IDC market research that SSD shipments increased 66% in Q3 with the biggest advances coming from Enterprise (storage and servers) and Client (PCs). Price increases in traditional hard drives will provide a boost to solid state drive adoption rates in consumer PCs and SANs deployed in enterprise and service provider clouds. Also the increasing volume of SSD shipments will be accompanied by lower SSD costs. Look for more SSD in the SANs.

---

Hybrid Cloud Computing
In-house solutions such as Cisco Flexpod, VCE, HP Cloud, Microsoft, and others will see accelerated growth as enterprises have seen the benefits of public Cloud Computing but prefer some of the comforts of in-house private Clouds. I expect growth in Hybrid Cloud Computing deployments.

With the Hybrid model specific aspects of IT infrastructure is moved to the cloud. This allows you to mix and match the resources between in-house infrastructure which is difficult to scale and cloud resources that’s scalable and can be provisioned on demand. An example would be a Business Intelligence application where the data is located in-house cloud and the processing is performed in the cloud.

---

Data Center Fabric Infrastructure Takes Hold
Cisco, Juniper, and Brocade have been touting the benefits of a data center “Fabric”. In 2012, I expect to see data center fabric networks, running over 10Gb Ethernet, to move from the “early adopter” phase to the “mass adoption” phase. (It could be argued that this is already happening).

Fibre Channel over Ethernet (FCoE) and Data Center Bridging (DCB) allows storage and network traffic to be transported across common lossless Ethernet links. Server adapters called Converged Network Adapters (CNA) unify storage and network I/O to FCoE data center switches.

Benefits of a Data Center Fabric infrastructure:

• Consolidated server I/O
• Reduced cabling
• Reduced power consumption
• Increased Virtual Machine mobility
• High-Speed, low latency interconnectivity
• Layer-2 connectivity

What do you think?

Home

Cisco Overtakes HP in Data Center Market

The global data center infrastructure hit US$26.2 billion in the third quarter of 2011, with Cisco Systems overtaking HP on aggregated revenue to take pole position, according to Canalys.

http://www.zdnetasia.com/cisco-overtakes-hp-in-data-center-market-62303222.htm

In a report released Wednesday, the research firm said worldwide data center infrastructure market grew 2.7 percent from US$25.5 billion in second-quarter 2011. Data center virtualization and consolidation, as part of efforts to migrate to private cloud, data center refresh and optimization project helped drive overall growth.

Industry standard servers remained the biggest part of the market in the third quarter, accounting for 39 percent of total investment, followed by storage at 25 percent, and Ethernet networking at 21 percent, Canalys said.

On an annual basis, industry standard server revenue increased by about 9 percent, but the strongest growth was in server virtualization which rose by approximately 30 percent.

Delving into vendor analysis, in the third quarter, Canalys said Cisco overtook HP to take pole position in the data center infrastructure market, followed by IBM in third position, Dell in fourth and EMC in fifth. In the x86 blade server market segment, Cisco is currently third but is expected to be second to HP by the end of 2012, the research firm predicted.

VMware ESXi 4.1 Optimizations for Cisco UC on UCS

In my previous post I discussed the importance of installing the latest version of VMware Tools on UC virtual servers in a Cisco UC on UCS deployment. A second and equally important VMware 'tweak' is disabling Large Receive Offload (LRO) on VMware ESXi hosts.

VMware Knowledge Base Article 1027511 addresses some Linux TCP/IP stacks perform poorly when handling LRO-generated packets. This results in poor TCP performance.

1. Log into the ESXi host or its vCenter with vSphere Client.
2. Select the host > Configuration > Software:Advanced Settings.
3. Select Net and scroll down slightly more than half way.
4. Set the following parameters from 1 to 0:
1. Net.VmxnetSwLROSL
2. Net.Vmxnet3SwLRO
3. Net.Vmxnet3HwLRO
4. Net.Vmxnet2SwLRO
5. Net.Vmxnet2HwLRO
5. Reboot the ESXi host to activate these changes.

Your guest VMs should now have normal TCP networking performance.

References:
http://docwiki.cisco.com/wiki/Disable_LRO
VMware Knowledge Base - KB Article 1027511
Cisco UC on UCS - Don't Forget VMware Tools

Home

Cisco UC on UCS - Don't Forget VMware Tools

Recently we upgraded a Cisco Unified Communications system running on physical HP servers, to Cisco UCS C-Series rack mount servers. Three days after the upgrade, all of the UC applications slowed down.

When users picked up their phone handset, there was a noticeable delay before dial tone. Delayed dial tone is a classic symptom of an overworked CallManager.

The Cisco UCS C210M2 servers were more than adequate for the load. Each UCS-C210M2 hosted VMware VMs of CallManager, Unity Connection, and UCCX. All of the VMs were deployed using the Cisco prescribed OVA templates.

Cisco UCS C210M2

So what is “VMware Tools”?

According to docwiki.cisco.com/wiki/VMware_Tools:

VMware Tools are specialized drivers for virtual hardware that is installed in the UC applications when they are running virtualized.

Now I understand Drivers. Up to date drivers are a good thing. Docwiki goes on to state:

It is very important that the VMware tools version running in the UC application be in sync with the version of ESXi being used.

Anyone who has read Cisco documentation before should realize, Cisco doesn’t often use the phrase “it is very important”. This is as close as Cisco gets to saying “you have to do this”.
So we installed VMware tools on all of the VMs and the Cisco voice system has been stable ever since.

Cisco has 3 methods for installing VMware Tools on Unified Communications Application VMs dependent on the UC Application version. Visit DocWiki-VMware Tools to find the appropriate method.

Please Note: VMware Tools will have to be re-installed anytime VMware ESXi is patched or upgraded.

Home

Virtual Desktop at My House?

As a typical geeky network engineer, I enjoy those times when the awesome technology I work with makes its way home. I remember in the late 90's introducing my wife to the joys of a dial up modem and the Internet. Years later I brought home a Linksys Wireless Router and Comcast Broadband Internet Connection. She wasn't that impressed when I said the cable Internet connection was faster than the T1 I used at work.

I need to preface this with, we have five girls, 10 months to 9yrs. They are very technologically advanced. For Christmas, the four oldest got Apple iPod iTouchs.

The girls like to use our laptops to go to nickjr.com and starfall.com. Ok, we can handle it, we have two laptops, we can all share. Oh No!!! It was World War III in our living room. Four girls, two laptops, "Daaadddd she has been playing for a long time!!"

By the time you read this, we are the proud owners of a 1/4 acre of land with a hole in it. We are building a house. My wife is doing a wonderful job designing the kitchen, selecting flooring, and picking paint colors.

I am working on the technology for the house. Fresh off the girls fighting for laptops, I wondered what could be done. Then the light bulb went off, Virtual Desktop! Why can't we have four Virtual Desktop terminals? I wouldn't have to worry about the kids deleting programs or documents, messing up configurations, and the static nature of having terminals, monitors, keyboard, and mice physically located in their own space.

So I am exploring my home based VDI initiative. I am not sure what I am getting into financially. I can piece together some physical servers with VMware. I need to find some terminals, possible Wyse devices.