I was under the impression from what I've seen talked about on this list (perhaps I'm wrong here) that given the write throughput of one node in a cluster (again assuming each node has a given throughput and the same config) that you would simply multiply that throughput by the number of nodes you had, giving you the total throughput for the entire ring (like you said this is somewhat artificial). The main benefit being that adding capacity was as simple as adding more nodes to the ring with no degradation.
So by your math, 100 nodes with each node getting 5k wps, I would assume the total capacity is 500k wps. But perhaps I've misunderstood some key concepts. Still a novice myself ;-) On Tue, May 11, 2010 at 3:08 PM, Bill de hOra <b...@dehora.net> wrote: > Mark Greene wrote: > >> If you have for example, your replication factor equal to the total amount >> of nodes in the ring, I suspect you will hit a brick wall pretty soon. >> > > Right :) So if we said there was 100 nodes at 5K wps with R=2, then would > that suggest the cluster can support 250K wps? > > Again, I know this is a tad artificial, just trying to understand the > impact of replication on writes. > > > > The biggest impact on your write performance will most likely be the >> consistency level of your writes. In other words, how many nodes you want to >> wait for before you acknowledge the write back to the client. >> > > I'd agree for any individual client; what I'm after is the overall capacity > a cluster has over time in the face of replicas. But let's assume it's > ConsistencyLevel.ONE - how would you think the available write capacity > degrades? > > Bill > > > >> >> On Tue, May 11, 2010 at 12:10 PM, Bill de hOra <b...@dehora.net <mailto: >> b...@dehora.net>> wrote: >> >> If I had 10 Cassandra nodes each with a write capacity of 5K per >> second and a replication factor of 2, would that mean the expected >> write capacity of the system would be ~25K writes per second because >> the nodes are also serving other nodes and not just clients? >> >> I know this is highly simplified take on things (ie no consideration >> for reads or quorum), I'm just trying to understand what the >> implication of replication is on write scalability. Intuitively it >> would seem actual write capacity is total write capacity divided by >> the replication factor. >> >> Bill >> >> >> >
