Underlying HCI infrastructure architecture considerations

As many organizations embrace digital transformation and the application modernization journey that is involved in this process, Dell Technologies and VMware supporting customers by providing them with modern cloud infrastructure and storage solutions that support the demands of this new set of cloud native applications.

Dell ObjectScale, VMware vSphere with Tanzu, and the vSAN Data Persistence Platform (vDPp) are all examples of next generation cloud native technologies that deliver simple, scalable, and enterprise grade Kubernetes native S3 compatible object storage services on a Kubernetes runtime built into the vSphere hypervisor. To learn more about the details of this powerful set of technologies, check out these great blog posts from my colleagues over at VMware here and here. A recently published reference architecture white paper also walks through the steps of deploying these technologies together. 

Now let’s get into our primary topic for this blog, which is the underlying HCI infrastructure architecture considerations for running ObjectScale on vSphere with Tanzu. 

Setting the stage

Cloud infrastructure administrators have a lot of flexibility in terms of what and how to configure the infrastructure on which Dell ObjectScale runs. These options not only come at the underlying HCI infrastructure implementation layer but also at the VMware SDDC layer. This gives administrators choices on mixing the right combination of the two layers that best meet their business and operational requirements.

So, what are the layers that make up these options? For this discussion we will break it down as follows:

HCI Infrastructure Layer Options

• Construct – Dell vSAN Ready Nodes
• Consume – Dell VxRail HCI Integrated Systems

VMware SDDC Software Layer Options

• 
  

  Construct - VMware vSphere with Tanzu + VMware NSX-T
• Consume - VMware Cloud Foundation (VCF) with Tanzu

After we review these options, we will highlight how they can be used to align to your ObjectScale architecture design and workload requirements.

Construct HCI and Construct VMware SDDC – Dell ObjectScale on Dell vSAN Ready Nodes with VMware vSphere with Tanzu + VMware NSX-T

This option involves deploying ObjectScale on vSphere with Tanzu enabled Dell vSAN Ready Node clusters and then manually deploying and configuring the rest of the required VMware SDDC software stack including NSX-T. This is essentially the builder’s approach to implementing the HCI infrastructure stack and the VMware SDDC stack. This gives infrastructure administrators the most control over their infrastructure configuration and components. The tradeoff, however, is that it adds a bit more complexity and more manual steps to get to an outcome that is ObjectScale ready.

Consume HCI and Construct VMware SDDC – Dell ObjectScale on Dell VxRail with VMware vSphere with Tanzu + VMware NSX-T

 With this approach, infrastructure administrators can take advantage of consuming pre-validated and co-engineered Dell VxRail HCI integrated systems, enabling vSphere with Tanzu on them, and then manually deploying the NSX-T components of the solution. This speeds up and simplifies the HCI infrastructure management and operations portion of the stack while still delivering on the required SDDC infrastructure foundations needed for ObjectScale to run.

Construct HCI and Consume VMware SDDC – Dell ObjectScale on VMware Cloud Foundation with Tanzu on Dell vSAN Ready Nodes

 This option delivers infrastructure administrators with granular control in constructing the underlying HCI HW components while simplifying the VMware SDDC layer and consuming it as a full cloud platform using VMware Cloud Foundation. This helps streamline the VMware SDDC to include NSX-T out of the box and can automate the deployment and configuration of the VMware SDDC components that are required to enable vSphere with Tanzu and run ObjectScale.

Consume HCI and Consume VMware SDDC – Dell ObjectScale on VMware Cloud Foundation with Tanzu on Dell VxRail

 This option provides a true full stack turnkey cloud infrastructure platform for infrastructure administrators to consume. This co-engineered solution between VMware and Dell Technologies delivers the fastest path to hybrid cloud and Kubernetes. Administrators gain the operational and feature benefits of VxRail, the only HCI system with deep VMware Cloud Foundation integration, with the out of the box simplicity and automation of the VMware Cloud Foundation SDDC cloud platform. From an ObjectScale use case perspective, infrastructure administators can accelerate getting all the needed underlying cloud infrastructure up and running so that ObjectScale can be deployed quickly and easily at scale and with a standardized cloud infrastructure architecture built in.

Choosing the right ObjectScale deployment infrastructure architecture

All these options deliver the necessary infrastructure prerequisites required to deploy and run ObjectScale, just through different implementation approaches that align to an organization’s operating model. ObjectScale, however, can also be deployed in several different ways, which can affect the implementation of your underlying infrastructure. 

 Let’s review what these options are, how our infrastructure can support these deployment models, and when would be the best time to choose one over the other.

 First, let’s call out the ObjectScale deployment architecture options available:

1. Co-locate ObjectScale data services on the same clusters where user application workloads run
2. Run ObjectScale data services on dedicated cluster infrastructure separate from user application workloads

How an infrastructure administrator would configure the underlying HCI and VMware SDDC stack based on these options will ultimately depend on which SDDC deployment method was used, vSphere with Tanzu + NSX-T or VCF with Tanzu.

The infrastructure implementation design details vary slightly since VCF implements a prescriptive cloud architecture using the concept of workload domains. This means that cloud infrastructure administrators must consider how to deploy vSphere with Tanzu enabled clusters to run ObjectScale within the context of this VCF’s workload domain architecture. On the other hand, if administrators were using the build approach of deploying individual vSphere with Tanzu enabled clusters, architecture design decisions are a bit more open ended. Either way, both implementation methods support both ObjectScale deployment architecture models of co-located and dedicated and can be run on both Dell vSAN Ready Nodes and Dell VxRail HCI Integrated Systems.

So, what would the first option look like when co-locating ObjectScale data services on the same cluster as where user application workloads are run?

The following figure provides a visual depiction of what this option may look like in a VCF on VxRail deployment using a single VI workload domain with a single vSphere with Tanzu enabled VxRail cluster in it. In this example, we would deploy ObjectScale to the Supervisor Cluster running on this WLD cluster. Application teams would then have their user application workloads running on the same cluster infrastructure and share the underlying physical HCI compute, network, and storage resources. 

 Figure 1: VCF on VxRail – ObjectScale co-location cluster deployment

This approach has advantages in terms of minimizing the infrastructure footprint required to run both workload types. It can also help drive improved resource utilization of the HCI infrastructure that has been deployed. This can also be a great fit for minimizing licensing costs if you have containerized user workloads and VM-based workloads that need to consume ObjectScale storage since there is only one cluster you need to enable vSphere with Tanzu on and vSphere can support running containers and VMs on the same vSphere cluster. However, there are possible downsides. These include resource contention for user workloads since you are sharing the same infrastructure to run ObjectScale data services and lack of independent scalability and right sizing of infrastructure resources for ObjectScale and the user applications.  

Option 2, running ObjectScale data services on dedicated cluster infrastructure separate from user application workloads, eliminates the resource contention by running ObjectScale on its own dedicated cluster infrastructure separate from user workloads. In a VCF on VxRail deployment, this may be implemented in a couple of ways. The first is to create a single VI WLD with two or more VxRail clusters in it. One cluster would have vSphere with Tanzu enabled on it and is where ObjectScale would be deployed. The other cluster, depending on the types of workloads running (whether they be VM-based only or a mix of containers and VMs) may not require vSphere with Tanzu be enabled on it and can just be used to run user application workloads.

By running ObjectScale on its own workload domain cluster resources, we now have physical resource isolation for both ObjectScale and user application workloads. This avoids resource contention between the two and now have the flexibility to independently scale resources for both as needed. Using this VCF workload domain organizational model may be helpful if your organization is aligning ObjectScale storage and the workloads that consume it as part of a single business unit and you may want to keep all of that together and managed within a single managed pool of cloud infrastructure resources. The following diagram provides an illustration of how this would look. 

 Figure 2: VCF on VxRail – ObjectScale dedicated cluster deployment with single VI WLD

The other VCF workload domain design approach is to deploy two VI workload domains. One would contain one or more VxRail clusters with vSphere with Tanzu enabled on them and ObjectScale would be deployed on top. The other VI workload domain would contain one or more VxRail clusters that may or may not have vSphere with Tanzu enabled on them and would run user application workloads only. This method still gets you separation of physical resources to avoid resource contention as well as independent scaling for both workload types, but organizationally we have deployed workload domains based on infrastructure service function. 

Deploying ObjectScale into its own dedicated workload domain provides the possibility of maximum scale of how many clusters we can deploy into a single domain that can be used solely for running ObjectScale data services. We can also help simplify the networking for those clusters since we only need to accommodate for the networking needs of ObjectScale and not also for user applications workloads, too. 

The following example uses dedicated NSX-T instances for each VI workload domain. In VCF, it is possible to share an NSX-T instance across multiple VI workload domains. If we would have done this, we wouldn’t have to deploy another cluster of NSX Edge appliances and could have just used the NSX Edge appliance deployed in VI Workload Domain 2 to meet the requirements that are needed when enabling vSphere with Tanzu on vSphere clusters. But since we are using separate dedicated NSX-T instances, each VI workload domain will require NSX Edge appliances to meet these vSphere with Tanzu and ObjectScale minimum requirements for the clusters contained within them. The following figure shows an illustration of what this multi-workload domain organizational model would look like. 

Figure 3: VCF on VxRail – ObjectScale dedicated cluster deployment with two VI WLDs

It is important to call out that these same co-located and dedicated cluster ObjectScale architecture models can be used in vSphere with Tanzu + NSX-T on Dell vSAN Ready Nodes/VxRail deployment options as well and are not tied to just the VCF on VxRail examples shown here. The same overall ObjectScale logical and physical layout considerations would apply. Administrators who choose to approach running ObjectScale in this way would be responsible for determining where the NSX-T Manager VM’s, Edge appliances, and vCenter components would run as there would be no Management Domain construct defined as part of a cloud platform architecture like VCF has.

This is not the end, it’s just the beginning…

I hope you have found this information helpful as you work through your ObjectScale adoption journey. This is not the end of your journey, however. For more information about VxRail and ObjectScale, check out the links at the bottom of this post.

Author: Jason Marques

Twitter: @vWhipperSnapper

Additional Resources

ObjectScale page on InfoHub

VxRail page on DellTechnologies.com

VxRail page on InfoHub

VCF on VxRail Interactive Demos

VxRail Videos

Quick! The word has just come down. There is a new initiative that requires a massively parallel processing (MPP) database, and you are in charge of implementing it. What are you going to do? Luckily, you know the answer. You also just discovered that the Dell PowerFlex Solutions team has you covered with a solutions guide for VMware Tanzu Greenplum.

What is in the solutions guide and how will it help with an MPP database? This blog provides the answer. We look at what Greenplum is and how to leverage Dell PowerFlex for both the storage and compute resources in Greenplum.

Infrastructure flexibility: PowerFlex

If you have read my other blogs or are familiar with PowerFlex, you know it has powerful transmorphic properties. For example, PowerFlex nodes sometimes function as both storage and compute, like hyperconverged infrastructure (HCI). At other times, PowerFlex functions as a storage-only (SO) node or a compute-only (CO) node. Even more interesting, these node types can be mixed and matched in the same environment to meet the needs of the organization and the workloads that they run.    

This transmorphic property of PowerFlex is helpful in a Greenplum deployment, especially with the configuration described in the solutions guide. Because the deployment is built on open-source PostgreSQL, it is optimized for the needs of an MPP database, like Greenplum. PowerFlex can deliver the compute performance necessary to support massive data IO with its CO nodes. The PowerFlex infrastructure can also support workloads running on CO nodes or nodes that combine compute and storage (hybrid nodes). By leveraging the malleable nature of PowerFlex, no additional silos are needed in the data center, and it may even help remove existing ones.

The architecture used in the solutions guide consists of 12 CO nodes and 10 SO nodes. The CO nodes have VMware ESXi installed on them, with Greenplum instances deployed on top. There are 10 segments and one director deployed for the Greenplum environment. The 12th CO node is used for redundancy.

The storage tier uses the 10 SO nodes to deliver 12 volumes backed by SSDs. This configuration creates a high speed, highly redundant storage system that is needed for Greenplum. Also, two protection domains are used to provide both primary and mirror storage for the Greenplum instances. Greenplum mirrors the volumes between those protection domains, adding an additional level of protection to the environment, as shown in the following figure:

By using this fluid and composable architecture, the components can be scaled independently of one another, allowing for storage to be increased either independently or together with compute. Administrators can use this configuration to optimize usage and deliver appropriate resources as needed without creating silos in the environment.  

Testing and validation with Greenplum: we have you covered

The solutions guide not only describes how to build a Greenplum environment, it also addresses testing, which many administrators want to perform before they finish a build. The guide covers performing basic validations with FIO and gpcheckperf. In the simplest terms, these tools ensure that IO, memory, and network performance are acceptable. The FIO tests that were run for the guide showed that the HBA was fully saturated, maximizing both read and write operations. The gpcheckperf testing showed a performance of 14,283.62 MB/sec for write workloads.

Wouldn’t you feel better if a Greenplum environment was tested with a real-world dataset? That is, taking it beyond just the minimum, maximum, and average numbers? The great news is that the architecture was tested that way! Our Dell Digital team has developed an internal test suite running static benchmarked data. This test suite is used at Dell Technologies across new Greenplum environments as the gold standard for new deployments.

In this test design, all the datasets and queries are static. This scenario allows for a consistent measurement of the environment from one run to the next. It also provides a baseline of an environment that can be used over time to see how its performance has changed -- for example, if the environment sped up or slowed down following a software update.

Massive performance with real data

So how did the architecture fare? It did very well! When 182 parallel complex queries were run simultaneously to stress the system, it took just under 12 minutes for the test to run. In that time, the environment had a read bandwidth of 40 GB/s and a write bandwidth of 10 GB/s. These results are using actual production-based queries from the Dell Digital team workload. These results are close to saturating the network bandwidth for the environment, which indicates that there are no storage bottlenecks.

The design covered in this solution guide goes beyond simply verifying that the environment can handle the workload; it also shows how the configuration can maintain performance during ongoing operations.

Maintaining performance with snapshots

One of the key areas that we tested was the impact of snapshots on performance. Snapshots are a frequent operation in data centers and are used to create test copies of data as well as a source for backups. For this reason, consider the impact of snapshots on MPP databases when looking at an environment, not just how fast the database performs when it is first deployed.

In our testing, we used the native snapshot capabilities of PowerFlex to measure the impact that snapshots have on performance. Using PowerFlex snapshots provides significant flexibility in data protection and cloning operations that are commonly performed in data centers.

We found that when the first storage-consistent snapshot of the database volumes was taken, the test took 45 seconds longer to complete than initial tests. This result was because it was the first snapshot of the volumes. Follow-on snapshots during testing resulted in minimal impact to the environment. This minimal impact is significant for MPP databases in which performance is important. (Of course, performance can vary with each deployment.)  

We hope that these findings help administrators who are building a Greenplum environment feel more at ease. You not only have a solution guide to refer to as you architect the environment, you can be confident that it was built on best-in-class infrastructure and validated using common testing tools and real-world queries.

The bottom line

Now that you know the assignment is coming to build an MPP database using VMware Tanzu Greenplum -- are you up to the challenge?

If you are, be sure to read the solution guide. If you need additional guidance on building your Greenplum environment on PowerFlex, be sure to reach out to your Dell representative.  

Resources

• Dell PowerFlex
• VMware Tanzu Greenplum

Authors:

• Tony Foster – Dell Technologies, Twitter: @wonder_nerd
  LinkedIn
• Sue Mosovich – VMware