Olaf Kock Vor 8 Jahren As you state that time is money: It's 23,28€ now (that's 11,28$ for the US guys, not sure how to calculate the "pm") and I shouldn't be commenting any more. However, it looks like clustering implies session replication, but it can very well run without it - and in most cases it should. The aspect clustering can't go without is cache invalidation (or cache synchronization) - the diagrams would be more accurate with this label. Do you agree? Melden Sie sich an, um auf diesen Beitrag zu antworten. Antworten als... Abbrechen James Min Olaf Kock Vor 8 Jahren Agree with Olaf. You could replace "session" with "cache replication/synchronization" in the diagrams. Melden Sie sich an, um auf diesen Beitrag zu antworten. Antworten als... Abbrechen Eric Kirby Olaf Kock Vor 8 Jahren Very good point, Olaf. The diagrams should now reflect this. Melden Sie sich an, um auf diesen Beitrag zu antworten. Antworten als... Abbrechen
James Min Olaf Kock Vor 8 Jahren Agree with Olaf. You could replace "session" with "cache replication/synchronization" in the diagrams. Melden Sie sich an, um auf diesen Beitrag zu antworten. Antworten als... Abbrechen
Eric Kirby Olaf Kock Vor 8 Jahren Very good point, Olaf. The diagrams should now reflect this. Melden Sie sich an, um auf diesen Beitrag zu antworten. Antworten als... Abbrechen
Mika Koivisto Vor 8 Jahren Java Pairing-Based Cryptography (JPBC) I think you were looking for JDBC. So we get HA nirvana with a single LB, I'd call that a single point of failure. What about running this thing in multiple data centers so that we don't have a single point of failure with the data center. I think that at least deserves a mention when talking about HA nirvana. Melden Sie sich an, um auf diesen Beitrag zu antworten. Antworten als... Abbrechen Eric Kirby Mika Koivisto Vor 8 Jahren Thank you for your feedback, Mika. You are correct, the diagrams should read "JDBC" as the connections highlighted in the diagram are between the application server and database. The diagrams should be updated to reflect this soon. However, the diagrams show a single data center for simplicity. Your point about the necessity of multiple data centers for high availability is completely valid and the article should now be updated to show this. Melden Sie sich an, um auf diesen Beitrag zu antworten. Antworten als... Abbrechen
Eric Kirby Mika Koivisto Vor 8 Jahren Thank you for your feedback, Mika. You are correct, the diagrams should read "JDBC" as the connections highlighted in the diagram are between the application server and database. The diagrams should be updated to reflect this soon. However, the diagrams show a single data center for simplicity. Your point about the necessity of multiple data centers for high availability is completely valid and the article should now be updated to show this. Melden Sie sich an, um auf diesen Beitrag zu antworten. Antworten als... Abbrechen
Krishna Pandey Vor 7 Jahren Very informative Blog Eric. I was looking around the implementation of the same solutions for our enterprise level application. We are going with Active - Passive mode. We believe we can achieve clustering through cluster manager of OS for this we are planning to purchase Red Hat Enterprise Linux OS. Melden Sie sich an, um auf diesen Beitrag zu antworten. Antworten als... Abbrechen