In case you couldn’t attend our Dell Cloud Chat with Lead Engineers: Cloud Optimized Servers & Blades please find below a summarized transcript. We cleaned it up and reordered some fragments, making the transcript more digestible. Our co-hosts in this chat:

Armando Acosta, Dell Product Manager. You can find detailed information about our Dell PowerEdge C Servers here.

Cedric Cook, Dell Platform Lead Systems Engineer. For detailed information about our Dell PowerEdge Blade Servers click here.

Why did Dell develop cloud servers?

DELL-Flo: @Armondo and Cedric: Why did Dell develop cloud servers?

Armando Acsota: Dell created DCS out of the unique needs of  hyperscale customers. Hyperscale includes cloud computing … cloud customers value different features. They value TCO, energy effciency and density. Cloud servers need a good balance of CPU, memory HD density, but this all needs to come in an efficient form factor. The more server per rack the better.

Cloud servers need to be efficient and enable cloud providers to do more work in a smaller amount of space.

What is the best way to match the account's business problem to the cloud optimized server and blade solutions that we have?

Tejas Gohokar: Well, from a marketing perspective, what is the best way to match the account's business problem to the cloud optimized server and blade solutions that we have?

Armando Acsota: Great question. It's all about customer need and what their infrastructure looks like. If a customer wants to run IaaS or PaaS they need larger resource pools that both blades and PE C can meet. But it depends how they want to deploy, maintain and scale in the future.

dell_cedric: I would like to add to what Armando stated. In that customers have different workloads that require different resources and infrastructure components, e.g. big pipes for bandwidth. Our PE line of Blades offer choice to accommodate those workloads, networking, and management requirements.

Armando Acsota: PE C is all about large scale deployments where customer need to get the highest density per rack and are willing to bring as much power to the rack to get it.

Trey Bohannon: I am hearing that we soon will be able to put 256 Gb of memory in the 6220. Is there a date for this?

Armando Acsota: C6220 supports up to 128GB today. 256Gb will be available in a day.

What is different about cloud servers compared to “regular” servers?

DELL-Flo: What is different about cloud servers compared to “regular” servers?

dell_cedric: "Regular Servers", I would classify as mainstream compute. I think that PE Modular platforms offer density, excellent power management features, and RAS. With the launch of 12G platforms, we've delivered: unprecedented node density with mainstream processor performance and memory scalability coupled in the industry’s most compact form factor.

… and speed the deployment of new blade servers while offering continuous “always on” management with the M420’s iDRAC7 with Lifecycle Controller “agent-free” embedded systems management. Dell’s robust “anytime, anywhere” management of platforms and infrastructure maximizes the ease of operation throughout the server lifecycle.

What is different in the management in the C servers vs. rack/blade?

Aldon Cheatham: Can you guys discuss management in the C servers vs. rack/blade etc and how the differences are meant to be that way etc.?

Armando Acsota: PE C management is down on a sled level through a dedicated BMC chip per server node that support IPMI 2.0. I've included a PE C management site that includes all the latest and greatest updates (see bottom of this page).

dell_cedric: Those PE Modular features are enabled by chassis management controller and iDRAC.

Armando Acsota: The page features quick start guides and deployment tool kits for Linux. You can now do one to many deployment on PE C with DTK (deployment tool kit).

Is it possible to use Intel System Management Software for monitoring and manging the PowerEdge-C Systems?

DELL-Flo: Is it possible to use Intel System Management Software for monitoring and manging the PowerEdge-C Systems?

Armando Acsota: You can use Intel node manager with PE C systems but you still need a console and integration point. We partnered with Joulex. They provide console and integration point with Intel Node Manager and PE C Server BIOS. Most customers use SNMP trap alerts to capture PE C BMC messages with an open source agent like Ganglia or Nagios. With C6220 there will be additional MIBs for better communication with SNMP. Check PE C managment website for all PE C management DTK and custom scripts. The page is updated on a weekly basis with ne winformation, it's very helpful for customers

Are customers seeing the value of end-to-end Dell solution which includes PE C and PE M?

dell_cedric: Question to you guys:  Are your customers seeing the value of end-to-end Dell solution which includes PE C and PE M ?

Armando Acsota: PE C has been very popular with Hadoop and OpenStack solutions.

dell_cedric: Ok, any takers on the previous question? Are your customers seeing the value of end-to-end Dell solution which includes PE C and PE M? Or do most customers adopt one over the other?

What are the advantages / disadvantages of M420 and PowerEdge-C and who are the target customers and infrastructures?

DELL-Flo: What are the advantages / disadvantages of M420 and PowerEdge-C and who are the target customers and infrastructures?

dell_cedric: So, PE M420 was explicitly designed to capture HPCC, rendering, web-farms, and can also accommodate virtualized environments

Armando Acsota: PE C is geared to 5 specific workloads HPC, hosting, big data, web 2.0, and cloud. The density that PEM 420 offers makes a great play towards environments that are foot-print conscious. PE C can be used in virtualization, but it doesn't have the interation into virtualized consoles like PE. PE has tighter into into vCenter … PE M420 also has individual serviceability.

PE C opportunity can be identified by listening to what customers are saying: “I am stranding power in my racks.” …  “We’re interested in running our data centers at higher ambient temperatures.” ... “We’re interested in running our data centers at higher ambient temperatures.” … “We developed our own systems management tools based upon IPMI.” … “We are tired of buying server-by-server and want to move to rack-level solutions.” … “We are using Nagios (or Ganglia) for our systems management today.”

Can a M420 be used in any M1000e Chassis or is there a limitation?

DELL-Flo: Can a M420 be used in any M1000e Chassis or is there a limitation?

dell_cedric: Good question: PE M420 was designed to integrate into the M1000e chassis as does the other 12G Blades. The intent for leveraging this common infrastructure makes PE M420 a great fit - World's most-dense Blade! I've posted a link to the spec-sheet for additional speeds- and feeds (see below).

With the C6220 we have Sled with a doubled height. Is it possible to use 2 HIC Cards to connect 16 GPU from a C410x?

DELL-Flo: With the C6220 we have Sled with a doubled height. Is it possible to use 2 HIC cards to connect 16 GPU from a C410x?

Armando Acsota: You can do 2 HIC cards in C6220 2 node, but max GPU toady per node is 4 GPUs. We are working on 8 GPU. New Intel architecture has created some challenges, but we are working through them. To be clear 8 GPU per node should be available in Nov / Dec timeframe.

What about operating system support? Can you tell us which OS are supported on PE-C and M420?

DELL-Flo: What about operating system support? Can you  tell us which OS are supported on PE-C and M420?

dell_cedric: Sure: PE M420 supports Windows 2008, RHEL 6.2, SLES 11, VMware, Citrix XenServer .. I'll post a link as well (see below).

Armando Acsota: PE C supports RHEL, SLES 11 SP1, Windows 2008, VMware, Citrix Xen.

Dell TechCenter: We have only 10 minutes left. Are there any further questions from the audience?

Will there be an updated C410x enclosure available soon, or is PCIe Gen3 fully supported with this?

Thomas: Will there be an updated C410x enclosure available soon, or is PCIe Gen3 fully supported with this?

Armando Acsota: At this time there is no plan of record for C410x follow on. If customers want a GPU and Gen3 solution, C8000 will be available soon.

I have received information about a possible redundancy of the NICs.  Can you provide more information on this?

dell_cedric: M420 design supports (3) redundant Fabrics.  A,B or C, and D.

Fabric A:  Dual-port Broadcom 57810s LOM device supporting 10GbE KR
Fabric B/C:  Quad-port and Dual-port Mezzanine Card supporting Fibre Channel, Ethernet, or Infiniband FDR10. I need to make one clarification… the Fabric B/C will also support Quad-port Mezzanine cards.
Fabric D: Supporting redundant Chassis Management

Below Fabric Mapping graphic describes how Fabric Connectivity works within M1000e chassis.  This image is also located in the PowerEdge M420 Technical guide (pg. 27) link that was posted.

Is it possible to use Qlogic infiniband mezzanine cards in a C6220?

Thomas: Is it possible to use Qlogic infiniband mezzanine cards in a C6220? (Customer ask because he want to use a C6000 Chassis with 4 Nodes and every Node should have a HIC).

DELL-Flo: No not yet. At the moment we have only Mellanox Connect X2 dual-port QDR IB and Mellanox Connect X3 single-port FDR IB as Mezzaniene Cards. This cards can be used together with Dell 410x HIC card.

Dell TechCenter: Thank you all very much for attending our chat

Dell TechCenter: @Cedric & @Armando Thank you very much guys for volunteering!

Resources
Dell PowerEdge C Series
Dell OpenManage Deployment
Dell m1000e Tech Guidebook
Dell OS Support
Intel Server Management
Intel Xeon e5 Workstation Integrated IO

Chat Hosts
Stephen Spector, Cloud Evangelist at Dell (Twitter: @SpectorAtDell)
Florian Klaffenbach, Solution Expert – Microsoft & Cloud Computing at Dell (Twitter:@FloKlaffenbach)
Rafael Knuth, Social Media Manager at Dell (Twitter: @RafaelKnuth)