Reading the first independent claim, from here on referred to as the claim, of US10630817B2 my understanding would be that they have patented any type of session-based load balancing, which includes for all sites using HTTPS where the TLS is offloaded to the load balancer.

But if one reads some of the figures and the abstract it's clear that they're referring to a destination options extension header, but in the claim, they are simply using the term header for the element which I then would assume to be any network message header.

My question is now, am I understanding this patent right, and isn't any website using HTTPS with offloading of the TLS handling to a load balancer infringing this patent?

I would think that the terms used as elements for the claim are way too broad. Also, session-based load balancing is most definitely prior art.

  • The linked document is an application. You will need to review the associated issued patent to see the approved claims.
    – Eric S
    Jul 1, 2022 at 17:51
  • patents.google.com/patent/US10630817B2/en Links to the patented version of this application.
    – George White
    Jul 1, 2022 at 22:58
  • Could you please refer to the linked patent from George White and edit your question based on the allowed claims?
    – Eric S
    Jul 3, 2022 at 21:29
  • Sure. That should be changed now.
    – techbech
    Jul 5, 2022 at 8:32

2 Answers 2


No, it does not apply to all session based load balancing.

I looked up the history of the application prosecution that resulted in this patent on Patent Center at the USPTO and went to the Global Dossier for the application.

I followed the rejection/amendment/rejection/acceptance of claim 1. The others had similar histories.

In the non-final rejection claim 1 was rejected as obvious under 103(a) under Ono 2003/0093560 in light of Ben-Nun 205/0190694.

The applicant responded with an amendment, adding the words

“determining based on the header not indicating the server for the first packet to be sent”

That overcame the rejection. The next rejection was a final rejection under 103(a) under two new references Kawata US 2002/0032777 and Achlicoptas US 2006/0233106.

The applicant prepared to appeal. The first step is to argue, on paper, to a small group within the examination corp. (not the actual appeals board) Usually its the examiner, their supervisor and a third primary examiner.

They argued that the feature that was amended in did not appear in either reference. The pre-appeal group agreed and returned it to examination and said wait for another action.

The next action was a Notice of Allowance. There was a telephonic discussion about claims 10-13 and a new NOA was issued. Fees paid and then patent issued.

You can see in the google patents record that some 15+ previous patent documents and a similar number of non-patent documents were looked at by the examiner in the course of prosecution.

By the way the same inventor has another patent that is specific to the destination options extension header but may not have other constraints these claim do have.

  • This is not to argue with you, I'm just confused as to why this claim was accepted. I would go to extend and say almost every form of session network load balancing is based on the idea a network packet being sent to a load balancer node, that determines which server the first message should be sent to based on a load balancing algorithm, and the load balancer then stores this decision with the source IP and port of that message, so when a message from the same source arrives the load balancer looks up the receiver of the previous message and then sends the new message to the same server.
    – techbech
    Jul 8, 2022 at 8:10
  • It might just be that I'm simply not understanding the language correctly, and I would really appreciate any explanation as to why this claim might not describe the same thing as I did in my previous comment.
    – techbech
    Jul 8, 2022 at 8:13
  • +1 for going through the effort of digesting the history of the patent prosecution.
    – Eric S
    Jul 8, 2022 at 15:26
  • 1
    The USPTO does it make this easy but you can read the winning argument the applicant made by downloading a 5 page pdf of the pre appeal brief submitted on 9-17-2019. I don’t think I can’t give a link that works because of the way public PAIR is set up. Go to public pair, put in the patent number without the US or the B2, go to the IFW tab. Find the document by the date I gave. They argue specifically why the distinguishing feature is not present in either of the two cited references.
    – George White
    Jul 8, 2022 at 20:09
  • Here is link for the PDF document @George is referring to.
    – techbech
    Jul 9, 2022 at 8:56

This really isn't my field of expertise. That said, a claim covers exactly what is stated, not more. To infringe on a claim, you must implement each and every element of the claim. If a claim consists of steps A, B, C, D and E and you only implement steps A, B, C and E, then you don't infringe on the claim. The first claim of the stated patent is as follows:

  1. A method comprising:

    receiving, by a load balancer, a first packet comprising a first header;

    determining, based on the first header not indicating a server identifier for the first packet to be sent and based on one or more load balancing criteria, a server from a plurality of servers;

    sending the first packet to the server;

    receiving a second packet comprising a second header;

    determining, based on the second header, that the second packet is associated with a same session as the first packet;

    and sending, based on a determination that the second packet is associated with the same session as the first packet, the second packet to the server.

The load balancer itself isn't what is claimed, that seems to be assumed as existing. What is claimed is a specific of instructions on how to process the packets based on the headers of both a first packet and a second packet.

As I said, this isn't my field, so I can't comment on how broad this is, but it does not seem to me to try to cover "all session-based network load balance?" as per you question's title. However, we would need to understand the other independent claims also (claims 10 and 14).

  • Thank you, Eric. This leave me with 2 questions. Do the order of the steps play any role? Let's say that a patent claim consists of A, B, C, I would go and invent a method with the same steps, but executed in order B, A, C. Would that infringe the claim? Also, you say we would need to understand the other independent claims. Do they somehow affect the validity of claim 1?
    – techbech
    Jul 8, 2022 at 7:38
  • 1
    @techbech It really depends on the claim. If the patent (and the claim) states the order is required than you could be okay. That said, claim language is a legal thing and I would consult with an attorney. Each independent claim (a claim that doesn't refer to another claim) is handled independently. So claim 1 could cover A and claim 10 could cover B. In principle there should be only one invention per patent, but there can be variations that are covered by different claims.
    – Eric S
    Jul 8, 2022 at 15:23

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .