Hyper-converged infrastructure based on server pairs

Patent metadata

Field	Value
Publication number	US-9778865-B1
Application number	US-201514847859-A
Country	US
Kind code	B1
Filing date	Sep 8, 2015
Priority date	Sep 8, 2015
Publication date	Oct 3, 2017
Grant date	Oct 3, 2017

Abstract

A hyper-converged infrastructure (HCI) provides one or more HCI units, each HCI unit including a pair of physical computing servers coupled to shared, nonvolatile storage. The shared, nonvolatile storage of each HCI unit is dedicated to that pair of physical computing servers and provides redundant storage of application data. Each of the pair of physical computing servers runs a set of application instances and an IO (Input/Output) stack. The application instances receive client requests over a network and generate IO requests specifying reads and writes of application data. The IO stack processes the IO requests to effect the specified reads and writes in the shared non-volatile storage.

First claim

What is claimed is:

1. A hyper-converged infrastructure (HCI) system, the HCI system comprising:
a pair of physical computing servers coupled to a network, each of the pair of physical computing servers configured to run a respective set of host application instances and to operate an IO (Input/Output) stack; and
a set of shared, nonvolatile storage devices coupled to each of the pair of physical computing servers and shared between the pair of physical computing servers, the set of shared, nonvolatile storage devices providing redundant storage of data of the host application instances,
wherein each of the pair of physical computing servers is constructed and arranged to (i) receive client requests over the network, the client requests directed to one of the set of host application instances running on that physical computing server, (ii) generate, via operation of that host application instance, IO requests specifying reads and writes of application data, and (iii) process, via the respective IO stack, the IO requests to effect the specified reads and writes of application data in the set of shared, nonvolatile storage devices,
wherein the IO stack running in each of the pair of physical computing servers exposes a respective set of host-application-accessible data objects, the set of host-application-accessible data objects built from the set of shared, nonvolatile storage devices, and
wherein the HCI system further comprises an administrative server constructed and arranged to provide a system-level view of host-application-accessible data objects across physical computing servers in the HCI system.

2. The HCI system of claim 1 ,
wherein the pair of physical computing servers includes a first physical computing server and a second physical computing server, and
wherein the HCI system further comprises a physical host computer coupled to the first physical computing server, the physical host computer constructed and arranged to generate, via operation of a host application instance running on the physical host computer, external IO requests specifying reads and writes of a host-application-accessible data object exposed by the IO stack of the first physical computing server, and
wherein the first physical computing server is constructed and arranged to process the external IO requests, via the IO stack of the first physical computing server, to effect the specified reads and writes of the host-application-accessible data object.

3. The HCI system of claim 2 , wherein each of the pair of physical computing servers includes multiple processing cores, wherein a first group of the multiple processing cores is dedicated to running the respective set of host application instances but not the IO stack, and wherein a second group of the multiple processing cores is dedicated to running the IO stack but none of the set of host application instances.

4. The HCI system of claim 2 ,
wherein each of the pair of physical computing servers further includes a cache, the cache including a set of volatile memory devices backed up by battery to maintain power to the set of volatile memory devices in the event of a power loss, and
wherein each of the pair of physical computing servers is further constructed and arranged to (i) store data specified in an IO write request, generated by a host application instance, in the cache of that physical computing server, (ii) mirror the data specified in the IO write request to the cache on the other physical computing server of the pair of physical computing servers, and (iii) acknowledge completion of the IO write request to the host application instance that generated the IO request after the caches on both physical computing servers have stored the data specified in the IO write request but before storing the data in the set of shared, nonvolatile storage devices.

5. The HCI system of claim 1 , wherein each of the pair of physical computing servers is further constructed and arranged to:
run a respective operating system (OS) virtualization framework;
run the IO stack of the respective physical computing server in a virtualized OS environment managed by the OS virtualization framework running on the respective physical computing server; and
run each of the set of host application instances of the respective physical computing server in its own virtualized OS environment managed by the OS virtualization framework running on the respective physical computing server.

6. The HCI system of claim 5 ,
wherein the OS virtualization framework includes a container-hosting component,
wherein each of the pair of physical computing servers is constructed and arranged to (i) run an OS kernel and (ii) run the container-hosting component on the OS kernel, and
wherein each virtualized OS environment is a respective container managed by the container-hosting component, each container being a software process that provides an isolated userspace execution environment running on the OS kernel.

7. The HCI system of claim 5 ,
wherein the OS virtualization framework includes a hypervisor for hosting virtual machines, and
wherein each virtualized OS environment is a respective virtual machine managed by the hypervisor.

8. The HCI system of claim 7 , further comprising a VM manager constructed and arranged to administer the virtual machines managed by each hypervisor running on the physical computing servers.

9. The HCI system of claim 8 , wherein each of the physical computing servers includes a component for communicating information about the virtual machines thereon to the VM manager.

10. The HCI system of claim 9 , wherein the administrative server is disposed within a virtual machine on one of the physical computing servers.

11. The HCI system of claim 8 , wherein the pair of physical computing servers and the set of shared, nonvolatile devices together form a first HCI unit, and wherein the HCI system further comprises:
at least one additional HCI unit, each additional HCI unit including its own pair of physical computing servers and its own set of shared, nonvolatile storage devices shared between that pair of physical storage servers, each physical computing server of each additional HCI unit having a hypervisor, a virtual machine constructed and arranged to run an IO stack, and another respective virtual machine for each host application instance run on that physical computing server; and
a network interconnecting the HCI units within the HCI system,
wherein the VM manager is further constructed and arranged to administer the virtual machines run on each additional HCI unit.

12. A hyper-converged infrastructure (HCI) system, the HCI system comprising:
multiple HCI units; and
a network interconnecting the HCI units,
wherein each HCI unit includes—
a pair of physical computing servers coupled to the network, each of the pair of physical computing servers configured to run a respective set of host application instances and to operate an IO (Input/Output) stack; and
a set of shared, nonvolatile storage devices coupled to each of the pair of physical computing servers and shared between the pair of physical computing servers, the set of shared, nonvolatile storage devices providing redundant storage of data of the host application instances,
wherein each of the pair of physical computing servers is constructed and arranged to (i) receive client requests over the network, the client requests directed to one of the set of host application instances running on that physical computing server, (ii) generate, via operation of that host application instance, IO requests specifying reads and writes of application data, and (iii) process, via the respecti

Assignees

• Emc Corp
• Emc Ip Holding Co Llc

Inventors

• Srinivasan Sudhir
• Reed Devon
• Cummins Daniel

Classifications

• G06F9/45558

  Hypervisor-specific management and integration aspects · CPC title

• G06F2009/45579

  I/O management, e.g. providing access to device drivers or storage · CPC title

• G06F3/0605

  by facilitating the interaction with a user or administrator · CPC title

• G06F3/0667

  at data level, e.g. file, record or object virtualisation · CPC title

• G06F3/0619 Primary

  in relation to data integrity, e.g. data losses, bit errors · CPC title