# US 2015/0149635 A1 · Method and system for distributed load balancing

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2015/0149635 A1
- Document ID: US20150149635A1
- Status: Published application
- Kind: A1
- Application number: 14/092,640
- Filed: 2013-11-27
- Published: 2015-05-28
- Earliest claimed: 2013-11-27
- Assignees: Avi Networks
- Applicants: Avi Networks
- Inventors: Ranganathan Rajagopalan; Murali Basavaiah; Kiron Haltore; Anand Parthasarthy; Abhijeet Joglekar
- Web page: https://patentgrep.com/patent/US20150149635A1

## CPC

- G06F 9/5083: {Techniques for rebalancing the load in a distributed system}
- G06F 9/45558: {Hypervisor-specific management and integration aspects}
- G06F 9/52: Program synchronisation; Mutual exclusion, e.g. by means of semaphores
- H04L 67/1014: based on the content of a request
- H04L 67/568: Storing data temporarily at an intermediate stage, e.g. caching
- G06F 2009/4557: {Distribution of virtual machine instances; Migration and load balancing}
- G06F 2009/45595: {Network integration; Enabling network access in virtual machine instances}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

Load balancing includes receiving, from a client, a connection request to establish a connection with a server; determining load balancing state information based at least in part on the connection request; synchronizing the determined load balancing state information across a plurality of service engines, including to invoke an atomic read-miss-create (RMC) function on a distributed data store service; and distributing the connection to a selected server among a plurality of servers according to a result of the RMC function.

## Claims

### Claim 1

A distributed load balancing system, comprising:
one or more processors configured to execute a plurality of service engines on the one or more processors, wherein a service engine of the plurality of service engines is configured to:
receive, from a client, a connection request to establish a connection with a server;
determine load balancing state information based at least in part on the connection request;
synchronize the determined load balancing state information across the plurality of service engines, including to invoke an atomic read-miss-create (RMC) function on a distributed data store service; and
distribute the connection to a selected server among a plurality of servers according to a result of the RMC function.

### Claim 2 (depends on 1)

The distributed load balancing system of claim 1, wherein the plurality of service is engines are configured to operate within a plurality of virtual machines executing on one or more physical devices.

### Claim 3 (depends on 1)

The distributed load balancing system of claim 1, wherein the load balancing state information includes one or more of: health information of at least some of the plurality of servers, persistence information associated with at least some of the connections, session information, client/server connection information, connection limitation information, or load balancing measurement.

### Claim 4 (depends on 1)

The distributed load balancing system of claim 1, wherein the distributed data store service is implemented using a database that supports the atomic RMC function.

### Claim 5 (depends on 1)

The distributed load balancing system of claim 1, wherein to synchronize the determined load balancing state information across the plurality of service engines includes to:
determine whether an entry corresponding to the load balancing state information exists in a data store;
in the event that no entry corresponding to the load balancing state information exists in the data store, create an entry corresponding to the load balancing state information in the data store.

### Claim 6 (depends on 5)

The distributed load balancing system of claim 5, wherein in the event that an entry corresponding to the load balancing state information exists in the data store, the service engine is further configured to determine whether the entry has expired.

### Claim 7 (depends on 6)

The distributed load balancing system of claim 6, wherein in the event that the entry has expired, the service engine is further configured to update a timestamp associated with the entry or to delete the entry.

### Claim 8 (depends on 1)

The distributed load balancing system of claim 1, wherein the service engine is further configured to:
check a local cache to determine whether an entry corresponding to the load balancing state information exists in the local cache; and
in the event that no entry corresponding to the load balancing state information exists in the local cache, determine whether an entry corresponding to the load balancing state information exists in the data store.

### Claim 9 (depends on 1)

The distributed load balancing system of claim 1, wherein the distributed data store service is implemented using a consensus-based distributed protocol.

### Claim 10 (depends on 1)

The distributed load balancing system of claim 1, wherein to synchronize the determined load balancing state information across the plurality of service engines includes to:
send the load balancing state information to a leader selected among a cluster comprising the plurality of service engines;
determine whether an entry corresponding to the load balancing state information exists in a global data store according to the leader; and
in the event that an entry does not exist in the global data store, add a new entry to the global data store and propagate new entry information to members of the cluster.

### Claim 11

A load balancing method, comprising:
receiving, from a client, a connection request to establish a connection with a server;
determining load balancing state information based at least in part on the connection request;
synchronizing, using one or more computer processors, the determined load balancing state information across a plurality of service engines, including to invoke an atomic read-miss-create (RMC) function on a distributed data store service; and
distributing the connection to a selected server among a plurality of servers according to a result of the RMC function.

### Claim 12 (depends on 11)

The method of claim 11, wherein the plurality of service engines are configured to operate within a plurality of virtual machines executing on one or more physical devices.

### Claim 13 (depends on 11)

The method of claim 11, wherein the load balancing state information includes one or more of: health information of at least some of the plurality of servers, persistence information associated with at least some of the connections, session information, client/server connection information, connection limitation information, or load balancing measurement.

### Claim 14 (depends on 11)

The method of claim 11, wherein the distributed data store service is implemented using is a database that supports the atomic RMC function.

### Claim 15 (depends on 11)

The method of claim 11, wherein synchronizing the determined load balancing state information across the plurality of service engines includes:
determining whether an entry corresponding to the load balancing state information exists in a data store;
in the event that no entry corresponding to the load balancing state information exists in the data store, creating an entry corresponding to the load balancing state information in the data store.

### Claim 16 (depends on 15)

The method of claim 15, wherein in the event that an entry corresponding to the load balancing state information exists in the data store, further comprising determining whether the entry has expired.

### Claim 17 (depends on 16)

The method of claim 16, wherein in the event that the entry has expired, further comprising updating a timestamp associated with the entry or to delete the entry.

### Claim 18 (depends on 11)

The method of claim 11, further comprising:
checking a local cache to determine whether an entry corresponding to the load balancing state information exists in the local cache; and
in the event that no entry corresponding to the load balancing state information exists in the local cache, determining whether an entry corresponding to the load balancing state information exists in the data store.

### Claim 19 (depends on 11)

The method of claim 11, wherein the distributed data store service is implemented using a consensus-based distributed protocol.

### Claim 20 (depends on 11)

The method of claim 11, wherein synchronizing the determined load balancing state information across the plurality of service engines includes:
sending the load balancing state information to a leader selected among a cluster comprising the plurality of service engines;
determining whether an entry corresponding to the load balancing state information exists in a global data store according to the leader; and
in the event that an entry does not exist in the global data store, adding a new entry to the global data store and propagating new entry information to members of the cluster.

### Claim 21

A computer program product for load balancing, the computer program product being embodied in a tangible computer readable storage medium and comprising computer instructions for:
receiving, from a client, a connection request to establish a connection with a server;
determining load balancing state information based at least in part on the connection request;
synchronizing the determined load balancing state information across a plurality of service engines, including to invoke an atomic read-miss-create (RMC) function on a distributed data store service; and
distributing the connection to a selected server among a plurality of servers according to a result of the RMC function.
