Does Dell have an answer for HPE Helion Openstack?
Apparently my representatives over at Dell didn't understand what I was looking for when I inquired via the old fashioned e-mail, and they pretty much punted. Fine. HP was happy to receive that ball and they have run with it. Football analogies aside I am still trying to understand exactly what Dell is offering here. In a nutshell here is what I want.
1. I want to be able to leverage Terraform to build my operating environments on a customer's premesis, or in the public cloud.
2. I want to separate myself from hardware support, and VMware support. I do not have the manpower to support the kinds of deployments I have used up to now (see my earlier posts to other forums).
3. I want a single point of contact for all things having to do with my deployed systems. If a piece of hardware goes bad, then I want Dell Support to know it, and schedule a time with me to replace it. i don't want to have to repeat the nightmare of trying to get an Equallogics controller into Africa and installed by remote hands while I work through nights fidgeting with this or that driver. I've had enough of that.
Hewlett Packard Enterprise seems to have an answer. Does Dell?
Just to be clear - you want this to be done by Dell itself and not partners - the reason I ask is that many organizations have sold their large services organizations in the last year (Dell, HPe, etc) and many of these functions are handled by partners. In the case of OpenStack Dell itself offers premanufactured/integrated versions of Openstack on their hardware (using Redhat and two other distributions) that is covered by the Dell support organization end to end. The VCE organization used to have an Openstack integrated solution based on vxRack for fully end-to-end supported environments - but found the Openstack community needed more flexibility and choice.
From my experience all of these solutions are better handed by a partner - who is better equipped to manage the day to day; can provide options for the physical location of such things; and can manage the global support experience. (Yes - I am biased because I run such a company :) )
What you describe above is actually the point of Virtustream and the VCE organizations;
The support level you mention with the proactive phone home is available on ALL Dell equipment using OME and Prosupport Plus support options.
Thank you Michael.
What kind of global footprint to these partners have? I have worked with IBM partners in the past who really failed to deliver what they promised. This is why I am looking for a large organization with global reach. Let's consider a rather off the beaten path city like Jakarta, Indonesia. What kind of day to day management can you provide for a system I might be depending on there? And if monitoring discovers a storage node has failed? How do you honor any hardware support agreement in such a place? I suspect an organization like Dell or HP would have some kind of local partnership to lean on in this case.
My experience with the Dell M1000e chassis is fairly extensive, and I have deployed them with M630s, MXLs, and other combinations of hardware in order to support the installation of VMware clusters for my own company's product to run on. Configuring every component of this system including the ESXi and VCenter is an extremely time consuming operation, likewise the licensing is not trivial. I want simplification, and reliability. I want Dell to design, and support a standard system that runs Openstack for me to use to deploy my software.
Am I looking for the DellEMC Ready Bundle. Or is this Jet-Pack?
This link offers some good sounding information >> blog.dellemc.com/en-us/jetpack-powered-openstack
Put an https:// in front of that one, and you will see this section:
Dell EMC PowerEdge FX Converged Architecture – Based on customer demand, this release introduces support for the PowerEdge FX-series converged modular server and storage platforms, providing greater choice and flexibility for Dell EMC’s OpenStack customers. In Release 10, we support the groundbreaking PowerEdge FX modular chassis along with the FC630 server for OpenStack compute and controller nodes, and the FC630 server with FD332 storage for massive direct attached storage capacity in easily scalable increments.
That sounds worthy, but what does this look like? How does it fit in a standard rack? How many pieces and parts are involved? I will keep looking.
Short Answer - the Ready Bundle is the preconfigured hardware/software bundle with extra sauce around RedHat OpenStack. JetPack is the technology that Dell has added to the Red Hat installation methodology that automates much of the process of deploying OpenStack to nodes in the configuration.
The FX Converged infrastructure will feel very familiar to you - its a 2U chassis that can take modified blades from the M1000e universe (they have FC in front of them vs M to differentiate) and adds in features like storage blades and PCIe slots. The switching in an FX2 is basically the MXL switch but smaller and with 10GbE uplinks.
If you want to see what an FX2 looks like you can see it at the Dell Virtual Rack
From the same web site you can look at the visuals on many of the Ready Bundles and options.
I havent really asked much about your workloads - some of what you are talking about might be possible with several different solutions. For example you might find that Enterprise Hybrid Cloud or Native Hybrid Cloud would deliver that experience you are looking for or Azure Stack might be something interesting...
Dont get me wrong - the RedHat bundle with CloudForms and OpenShift is excellent because it layers on top of native OpenStack management of public and private cloud, traditional mode one virtualization systems like VMware and DevOps enablement around containers.
I have asked one of the Dell OpenStack folks to join this conversation - hopefully he will post :)
In terms of support and using partners like Mavenspire - well we absolutely rely on the global logistics of Dell to ensure parts and hands are available around the world within the hardware SLA. For most of the software work physical presence is not required - but occassionally we will have to work through organizations like the TrustX Alliance - which expands our resources 100 fold. My engineers do travel around the world every few years aggregating customer projects together - in most of our cloud architectures we are working with two kinds of workloads - the cloud native (micro-service and ephemeral app) environments most are done at scale do individual components are not really that important and do not require the same level of 4 hour response that we might demand for traditional applications that rely on hardware for redundancy. That being said, for traditional applications there are lots of ways to architect to remove single points of failure in hardware - although sometimes you will see a blip as HA does its work.
We have supported systems as far away as Nepal and as close as around the corner - physical presence is to be minimized whenever possible.
Here is my drawing based on the documentation of the recommendation in this here Ready Bundle Architecture guide. This is very interesting, and a completely different strategy than HPE is using. I look forward to more information about this, and some clarification.
The drawing looks a little off - but it depends on what you are trying to do. There is a picture of the 4 FX2 configuration in the bundle the page after the text description.
Also note that the guide is written based on 13G hardware - the new 14G hardware is just coming out and with it the change to 25/100GbE networking. Its backwards compatible so it can take either 25/10 or 100/40 connections.
PS: One of the reasons choice looms large in open stack environments is that some people like to mix the FX and R solution bundles for various reasons. The R series guide hits other options such as external SAN options, multi-rack expansion and so forth.
Consider that the three Ceph nodes in the picture above will be limited to the 16 2.5inch disks in the storage sleds. Using 3 rack server (R730xd/R740xd) provides much larger capacity disk options with larger number of disks slots - again it just depends on what you are trying to do.
At this point I will introduce some requirements. If you will allow me to make the comparison of the Openstack "Project" to the the AWS "Virtual Private Cloud" (VPC). I know this is not a direct equivalency, but roughly speaking. I intend to have a stack of machines installed by DELL (or a partner) that will allow me to Terrraform a Project that will be used by my clients. The Project and the instances will be my responsibility. All of the hardware and software used by the Openstack (Storage, Compute, Controller), cabling PDUs, power cords everything will be monitored and managed by DELL (or a partner). For my Project I will require:
Stay tuned for more detail.
This is my product in essence. The stacks on the bottom of the picture will feed data to the nodes on the prod network. The entire prod stack will run on Openstack.
The tDB instances need storage that can handle >10K IOPs. The other instances are ok with standard IOPs <3k for instance.
Capacity is dependent on the clients volume. My standard images for these instances are as follows.
Given that I have 3 client stacks here the numbers for high IOPs storage will be tripled. Therefore as a beginning point I have the following requirements:
high IOPs storage: 2.4 TB Standard IOPs: 420GB
Processor Cores: 22 RAM: 58G
After deployment and client ramp up I will likely need to increase all of these numbers, and I believe Openstack gives me the perfect platform to expand. As my needs increase I expect my vendor to be able to expand this stack without requiring downtime. Does the Dell Opestack provide this kind of functionality?