White Papers

Predeployment
8 Dell EMC SC Series: Microsoft Exchange Server Best Practices | CML1037
excessive memory consumption by the Microsoft Exchange Replication service
(MSExchangeRepl.exe), and also for severe cases in which the system is in such a bad state that
threads cannot be scheduled.
• DAG lagged copy enhancements: Lagged copies can care for themselves to a certain extent using
automatic log play down. In addition, lagged copies can leverage Safety.Net (previously Transport
Dumpster in Exchange 2010), making recovery or activation much easier.
1.3.2 Exchange 2019
Exchange 2019 builds upon the performance enhancements of Exchange 2016 with a number of additional
enhancements. The changes for Exchange 2019 are mainly focused around security, but there are some
storage changes designed to improve performance and scale. These changes can reduce I/O usage, lower
latency, and improve scale — all at the same time.
• Tiered storage: The biggest change is around tiered storage. Exchange now can dual-write to SSD
and spinning media. This allows Exchange to access more frequently used data on faster, lower
latency media. This is due to the new Metacache Database implementation. The performance
improvements increase user-per-server density and lower I/O latency.
• Search indexes: A new search index architecture also improves the quality and performance of
search indexes. To increase performance for users, the index is stored per mailbox. This reduces the
index sizes and reduces the number of rebuilds required. The overall result is fewer index rebuilds for
users and smaller indexes, requiring less storage performance.
• CPU utilization: The core count Exchange can leverage has increased to 48 cores in Exchange
2019. This supports more parallel tasks and can increase the number of I/O streams a server can
generate based on the workload it is processing. This needs to be considered when designing
storage pools.
• Dynamic database cache: The goal of the Dynamic Cache is to tune memory more effectively for
active and passive databases. This results in more memory for active databases, increasing the
cache read hits. This will result in fewer reads.