The Next Multi-Billion Dollar Market For Service Providers A profitable answer to public cloud

Transcription

1 The Next Multi-Billion Dollar Market For Service Providers A profitable answer to public cloud White Paper trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies.

2 Introduction Traditionally, MSPs have worked with standard business applications like IT helpdesk, managed storage, backup, disaster recovery, , managed VoIP, telco services, remote monitoring, patch management and even virtual CIO services. While this is still a huge market (according to Everest Group, the MSP market globally is around $85-90 billion), many MSPs are concerned about the growing popularity of public clouds and the shrinking footprint of standard enterprise applications. There are two key reasons for these concerns: Standard Enterprise Applications Are Converting to SaaS All of the major SaaS vendors are offering one or more of these large enterprise applications as a service. Salesforce, Zoho, and SugarCRM offer CRM; NetSuite offers a financial application suite, Workday offers HR and employee management, and service is moving to Office 365 and Gmail. So, the footprint of these large, complex, expensive, mission-critical applications for enterprises is shrinking. Any modern-day company will have six to eight SaaS contracts for all these business applications, and they don t need anyone to help run these applications; most of the new SaaS companies offer consulting, and there is no great value that MSPs can offer for such applications. Workloads Moving to a Pay-As-You-Go Consumption Model As customers are looking for platforms to run workloads, they are moving to large public cloud vendors and looking to deploy workloads on those platforms. The ease of use, transparent pricing, and quick pay-asyou-go consumption make it attractive for companies to try public clouds and often use them as a benchmark for their next-generation consumption model. Once a workload moves to public cloud, the associated backup, disaster recovery, performance monitoring and management are also moving to that new platform. A lot of revenue is lost for MSPs due to this move. The Next Big Market for MSPs and Cloud Providers Based on Mark Hurd s Oracle Open World keynote in 2018, the GDP growth in various sectors and countries is around two percent. This reflects that the overall IT budgets are not growing a lot over time, and the demand for standard enterprise applications is not growing quickly. However, the growth in revenue of cloud companies is almost 40 to 60 percent. MSPs need to offer a platform ideal for building, testing and production deployments of custom applications of all types. In most cases, the workloads running in the cloud are custom applications. These can fall into different categories like SaaS applications, web applications, mobile back-ends, big data analysis, business intelligence, security and machine learning, but basically, cloud has become the de facto standard for building and deploying all kinds of new applications. These applications represent the next big market opportunity for service providers. MSPs offer a platform to build, test and do production deployment of custom applications of all types. In fact, public clouds have enabled the creation of cloud as a standard platform, validated a consumption model and also made it very popular for enterprises to outsource their infrastructure needs to someone else. Most enterprises don t want to deal with infrastructure, compute, storage, etc., and they just want to consume these in a self-service manner. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 2

3 What MSPs Must Do Cloud providers cannot win in this market using the standard ways of building infrastructure. The standard way of using servers, storage, networking, virtualization software stack, monitoring and operations software from different vendors and stitching it together with custom scripts is not going to scale. That will not be cost-competitive in this market. Rather, there are two key ingredients MSPs must offer in their hosting platform:» Public cloud economics» Integration with public clouds to offer choice and flexibility to customers A cloud-based infrastructure typically consists of the following components:» Software-defined compute» Software-defined storage» Software-defined networking» Cloud operations - monitoring, upgrades, events, alarms, capacity planning» Orchestration software for application deployment, provisioning» Showback, chargeback for billing, cost management These components typically require stitching together multiple software and hardware solutions along with expertise in one or more software stacks to build, operate and manage the cloud. For example, one can get servers from Lenovo/HPE/Dell, storage from Pure storage/nimble storage/netapp/emc, software stacks such as vcloud software suite from VMware, and then hire a few VMware experts and create a cloud solution. After doing all this heavy lifting, MSPs will finally get a cloud infrastructure they can offer to customers for consumption. In many cases, MSPs will have to create an application portal on top to offer an application as a service instead of just infrastructure. The three main costs in a cloud are hardware, software, and operations. Different technologies address the cost and complexity at these three layers. For example, hyper-convergence addresses the hardware cost by converging compute and storage and using standard x86 servers with local disks attached to them. Vendors like VMware (vcloud), Microsoft (Azure pack) and Red Hat (OpenStack) provide software suites to build a cloud using any standard hardware. These software suites convert a set of servers into a cloud. In addition, there are a lot of vendors providing monitoring and management solutions to help with operating a cloud environment. One can argue about which set of solutions is best and does the most in lowering the cost and complexity of building, managing, operating a cloud, but the best cloud is one where the MSP doesn t even have to worry about building, managing and operating it at all. This is the ZeroStack value proposition. The ZeroStack Software-Managed Cloud ZeroStack offers a cloud solution that converts any set of industry-standard servers with local disks into a unified cloud infrastructure. The solution consists of two software components: (1) an operating system called the ZeroStack Cloud Operating System (Z-COS) that installs on servers and (2) a cloud management application running as a web service (Z-Brain). The operating system consists of a hypervisor, cloud software and a highly resilient control plane that knows how to pool all the resources across servers, bring up all the services and monitor the cluster for any software or hardware failures. Telemetry from this cluster is reported to Z-Brain for analytics, machine learning and improving efficiency of the cloud. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 3

4 Proactive Remediation Data Processing Cluster Health Monitoring Events Stats Telemetry Figure 1: ZeroStack s Z-Brain architecture. ZeroStack s intelligent software dramatically reduces operational tasks, and learns by leveraging a big-data layer that stores and analyzes rich telemetry using machine learning taking the guesswork out of capacity planning, upgrades, ongoing management and troubleshooting. By combining an on-premises, distributed control plane with web-based monitoring and analytics, ZeroStack offers a cloud to customers that is completely managed by its software stack and support. There is no need for a customer to understand how a cloud is built. Operations like upgrades, patching and failure handling are all done by our software. There are several other features that are built into the platform such as VM high availability, reliable storage and data management, placement policies, an App store, cost management and GPU resource allocation. The figure below shows the overall architecture and some of the key capabilities. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 4

5 Cloud-based monitoring & operations IT Admin & Developers UI/API On-Premises Secure Connection Hyper-converged Infrastructure & management software with self-healing design CERTIFIED Figure 2: ZeroStack s cloud software converts a cluster of servers into a web-managed private cloud. This on premises cloud provides self-service consumption, monitoring, and an integrated learning engine for performance and efficiency insight. ZeroStack enables multitenant, multi-cloud, and containerized environments with a future-ready architecture. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 5

6 A Sampling of Services Available from ZeroStack s AppStore GPU-as-a-Service Artificial Intelligence and Machine Learning products and solutions are quickly becoming commonplace and are shaping our experiences in computing like no other time in history. Interactive speech (e.g. Alexa, Google Home, etc.), Visual Search and recommendation engines are just a few of the consumer applications that are available today on our phones, websites and e-commerce platforms. The impact of machine learning is getting broader with enterprise applications in health sciences (e.g. Dr. Watson), finance, security, data centers and cyber surveillance. General-purpose CPUs cannot deliver the user responsiveness and inference latency required by complex deep learning and AI workloads. That s because unlike GPUs built for this purpose general purpose CPUs are not designed to rapidly perform parallel operations on large amounts of data, e.g., multiplying matrices of tens or hundreds of thousands of numbers. Processing large data sets through the same hypothesized algorithm for learning and for intelligent inference is a fairly common operation in machine learning and deep learning applications. ZeroStack automatically detects and displays supported GPUs available in the cluster. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 6

7 Project member users can enable GPU acceleration and choose one or more GPUs to attach to their workload. ZeroStack s GPU-as-a-Service capability gives customers powerful features to automatically detect GPUs and make them available in the ZeroStack environment. In order to maximize utilization of this powerful resource, cloud admins can configure, scale, and allow fine-grained access control of GPU resources to end users. Users can enable GPU acceleration, deploy new machine learning and deep learning workloads with tools such as TensorFlow, Caffe, etc., and provide the apps dedicated access to multiple GPU resources for an order of magnitude, faster inference latency and user responsiveness. ZEROSTACK GPU-AS-A-SERVICE: Powerful and Simple Automated detection of supported GPU cards available in the cluster Built-in governance and fine-grained access control of GPU resources Self-service Ready ZeroStack automatically performs PCI scans to retrieve GPU inventory in the system. Cloud administrators can control which business units have access to which GPU resources. ZeroStack GPU-as-a-service is easy to use once access is provided by the cloud admins. Users enable GPU acceleration and attach GPUs from the available list to their workloads including Windows, Linux, and CUDA library support. Users can just as easily detach the GPUs once they are done using them. Built-in production operational capabilities For field maintenance, administrators can add and remove GPUs on existing servers by following host evacuation workflow best practices. The ZeroStack cluster can be scaled up on demand by adding new physical nodes to the cluster with GPU resources. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 7

8 ZEROSTACK GPU-AS-A-SERVICE: Use Cases MSPs who want to provide a GPU service to their end customers Deep learning and Machine Learning applications Agile Operations Automated detection of GPU resources, self-service option for end users, controlling customer access to GPUs, as well as cost management make ZeroStack GPU-as-a-service an attractive new revenue source for MSPs. Multiple GPU capabilities, dedicated/full GPU access, and self-service make this attractive for enterprises and universities who want to provide on-demand access to their users. ZeroStack s self-healing features, software-defined networking and storage as well as seamless upgrades and scalability can now be used with GPU applications as well. 1. CPU: Intel Xeon E2630 v4 10 core processor w/ virtualization (VT-x) and IOMMU (VT-d) support or a similar AMD CPU with AMD-V and AMD-Vi support 2. RAM: Minimum 80 GB DDR MHz (128 Gb recommended for Deep learning apps) 3. Storage: At least 2 TB HDD (7200 RPM) + 1TB SSD 1. Windows 2012, 2014 server Operating Systems 2. RHEL, CentOS, Ubuntu OS 3. CUDA libraries Supported GPUs NVIDIA TESLA GPU Cards Tesla P100 delivers superior performance for HPC and hyperscale workloads. Based on PASCAL architecture, it supports more than 21 teraflops of 16-bit floating-point (FP16) performance. NVIDIA P100 trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 8

9 Supported GPUs NVIDIA TESLA GPU Cards The GeForce GTX 1080 Ti is NVIDIA s new flagship gaming GPU, based on the NVIDIA Pascal architecture. It s equipped with next-gen 11 Gbps GDDR5X memory and a massive 11 GB frame buffer. NVIDIA GTX 1080 Ti DevOps-as-a-service DevOps has become an essential best practice and digital transformation tool to enable agility and flexibility for many organizations today for whom modern agile software delivery is a key differentiator. DevOps is a set of tools, processes, and automation with a focus on breaking down silos between development, operations, and infrastructure in order to improve velocity of product delivery across the entire lifecycle from concept to production. Pulling these teams together will require a cultural transition, internal champions, flexible API-driven infrastructure, and the tools necessary to facilitate an organization s digital transformation. The above diagram is a representation of the continuous iteration, continuous delivery methodology DevOps seeks to achieve. This requires multiple tools for various phases that need to be deployed, configured, integrated and maintained. ZeroStack takes care of all this heavy lifting so developers can focus on building and delivering their applications faster. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 9

10 ZeroStack s DevOps-as-a-Service gives software development teams an easy way to consume on-demand compute, storage and networking resources as well as provide easy access to continuous integration and continuous deployment (CI/CD) tooling and software services that help increase their development throughput and shorten product delivery. Additionally, it empowers operations teams to manage, maintain and operate the entire infrastructure environment with very few people using smart software that drives automation and intelligence into the entire stack. ZEROSTACK DEVOPS-AS-A-SERVICE: Self-Service DevOps Tools with IT Control Built-in AppStore with a variety of DevOps tools and applications The built-in AppStore offers dozens of application templates that enable customers to deploy DevOps tools and applications with ease. Some of the example templates include the following; CI/CD tools such as Jenkins, CloudBees, GitLab, Puppet, Ansible, Apache Maven and GoCD SQL and NoSQL databases such as Cassandra, Redis and MongoDB Dev Tools and Language stacks such as LAMP stack, MEAN Stack, DevBox, Terraform and Memcached Container tools such as Kubernetes, OpenShift and Docker Cloud administrators can control which business units have access to which apps while specifying usage quotas for storage, networking and CPU resources Built-in governance and fine-grained access control of underlying infrastructure resources trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 10

11 ZEROSTACK DEVOPS-AS-A-SERVICE: Self-Service DevOps Tools with IT Control (cont.) ZeroStack DevOps-as-a-Service is easy to use once access is provided by the cloud admins. Developers deploy their own apps from the AppStore, network distributed apps using the built-in software-defined networking, monitor usage and plan capacity, and troubleshoot and correlate issues using the built-in tools Self-service ready Built-in production operational capabilities For field maintenance, administrators can add and remove physical hosts following host evacuation workflow best practices. The ZeroStack cluster can be scaled up on demand by adding new physical nodes to the cluster. ZEROSTACK DEVOPS-AS-A-SERVICE: Use Cases MSPs who want to provide a DevOps-as-a-Service to their end customers Automated deployment of curated open source tools as well as commercially supported tools, self-service option for developers, controlling customer access to infrastructure resources as well as cost management make ZeroStack DevOps-as-a-Service an attractive new revenue source for MSPs. Enabling CI/CD, ML, IoT, and many more emerging technologies Multiple GPU capabilities, dedicated/ full GPU access, streaming analytics tools, small footprint and self-service make this attractive for enterprises and universities who want to develop IoT and ML applications in addition to CI/CD pipelines. Agile Operations ZeroStack s self-healing features, software-defined networking and storage as well as seamless upgrades and scalability can now be used with GPU applications. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 11

12 Big Data-As-A-Service The volume, velocity, and variety of data being generated today has overwhelmed the capabilities of current infrastructure and analytics solutions. We are now experiencing Moore s law for data growth: data is doubling every 18 months. IDC forecasts that by 2025, the global datasphere will grow to 163 zettabytes (that is a trillion gigabytes) that s ten times the data generated in In order to solve this data explosion challenge, modern Big Data solutions must use a slew of new tools such as Hadoop, Spark, Tensor Flow (for deep learning) and NoSQL databases such as MongoDB, Cassandra, etc., which all have to be deployed, integrated, and operated as a whole. Additionally, these use cases have widely varying demands on the underlying infrastructure for performance, latency, and capacity requirements. Operating this diverse set of requirements manually can be daunting, time-consuming, and expensive. ZeroStack s Big Data-as-a-Service capability gives customers powerful features to automate deployment, integration, and operations of a variety of popular and modern Big Data tools, providing self-service options to data scientists and developers while allowing the cloud administrator to maintain control via quota management and fine-grained access control. Users have access to several Big Data tools they can deploy with a few clicks into their private workbench environment. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 12

13 ZEROSTACK BIG DATA-AS-A-SERVICE: Self-Service Big Data Tools with IT Control Built-in App Store with a variety of Big-Data applications Built-in governance and fine-grained access control of underlying infrastructure resources The built-in App Store offers pre-built application templates that enable customers to deploy Big Data applications with ease. Some of the example templates include the following; Big Data applications such as Apache Hadoop, Cloudera Express, and Spark SQL and NoSQL databases such as Cassandra, Redis, MongoDB Monitoring and data analysis tools such as ELK, Splunk Application servers such as Apache and Nginx Container tools such as Kubernetes and Docker Cloud administrators can control which business units have access to which apps while specifying usage quotas for storage, networking, and CPU resources Self-service ready ZeroStack Big Data-as-a-Service is easy to use once access is provided by the cloud admins. Users deploy their own apps from the app store, network distributed apps using the built-in software-defined networking, monitor usage and plan capacity, troubleshoot and correlate issues using the built-in tools Built-in production operational capabilities For field maintenance, administrators can add and remove physical hosts following host evacuation workflow best practices. The ZeroStack cluster can be scaled up on demand by adding new physical nodes to the cluster ZEROSTACK BIG DATA-AS-A-SERVICE Use Cases MSPs who want to provide a Big Data service to their end customers IoT and Machine Learning applications Agile Operations Automated deployment of Big Data tools, self-service option for end users, controlling customer access to infrastructure resources as well as cost management make ZeroStack Big Data-as-a-Service an attractive new revenue source for MSPs. Multiple GPU capabilities, dedicated/ full GPU access, streaming analytics tools, small footprint, and self-service make this attractive for enterprises and universities who want to develop IoT and ML applications. ZeroStack s self-healing features, software-defined networking and storage as well as seamless upgrades and scalability can now be used with GPU applications as well. Customized services are the new path to profit for MSPs, who are in a unique position to combine their deep understanding of the customer s needs with the ability to respond to those needs. However, traditional infrastructure is too complex, expensive and time-consuming to deliver the service agility MSPs need. ZeroStack s Software-Managed cloud makes it easy and costeffective to deploy a cloud platform, and our service templates enable rapid rollouts of profitable customized services. trademark of ZeroStack, Inc. in the United States and/or other jurisdictions. All other marks and names mentioned herein may be trademarks of their respective companies. 13