vvern

10 ms·

61.
by vvern 4y ago
Here's two reasons which come to my mind: * Operationally, there's no turn-key, hosted solution. Getting it set up locally for testing and CI is not trivial. You have to really want to use it. This criticism 100% applies to Anna t
62.
by vvern 5y ago
That’s not quite right. You can very much use the bsl code in production without a license and many, many companies do. The BSL contains a space for additional use grants. In crdb’s license, you can use the product for anything except build
63.
by vvern 5y ago
One of these is not like the others.
64.
by vvern 5y ago
It absolutely does. What it does not support is sub-table level partitioning and fine-grained replication controls.
65.
by vvern 5y ago
Expression-based indexes were added in the previous release. Partial indexes have been around for two releases. Those two combine pretty well. Temp tables exist and have existed for two years, but the implementation is not very good; they’r
66.
by vvern 5y ago
Do you have benchmarks to back that up? The efficiency per core is worse, but I’ve not seen 10x worse. One thing to be careful of is making sure to have parallelism in your benchmark. Single-threaded Postgres may well be 10x faster on workl
67.
by vvern 5y ago
There are advantages to embedding. You can retain the host language type system and and object model. If you have a great query language and model but have to write a ton of code to marshal back and forth, it might not be adding that much v
68.
by vvern 5y ago
What does the interop with go look like? Is there a story for writing prolog queries over go data structures and iterating the results. If this could be made ergonomic, it’d be very cool.
69.
by vvern 5y ago
Deques are great. I think there's two ways I'd consider designing this library differently: 1) The resizing seems worse to me than a linked-list of fixed-size ring buffers which use a sync.Pool. 2) (more out there) some of the tim
70.
by vvern 5y ago
You know it. SERIALIZABLE + no stale reads. https://www.cockroachlabs.com/blog/consistency-model/
71.
by vvern 5y ago
Weird to not see any references to https://github.com/golang/go/issues/48409 / https://github.com/golang/proposal/blob/master/design/48409-... There's been
72.
by vvern 5y ago
citus has some rather severe limitations around foreign keys. Namely, you can’t use them from any distributed tables. See https://docs.citusdata.com/en/v10.2/develop/reference_ddl.ht...
73.
by vvern 5y ago
update: they missed implementing this and now go1.17 will be required to bootstrap go1.19+. https://github.com/golang/go/issues/44505#issuecomment-94259...
74.
by vvern 5y ago
Versions after go1.4 up to 1.17 need to be built with 1.4. From 1.18 on, they need to be built with 1.16. See https://github.com/golang/go/issues/44505 .
75.
by vvern 5y ago
8. Online upgrades.
76.
by vvern 5y ago
Is it using https://github.com/tvondra/tdigest under the hood, or a separate implementation?
77.
by vvern 5y ago
Temporary tables in cockroach exist, but the implementation was done largely to fulfill compatibility rather than for serious use. The implementation effectively just creates real tables that get cleaned up; they have all the same durabilit
78.
by vvern 5y ago
CockroachDB dev here. We've gotten a good bit better in the last few versions in terms of schema change stability. We're still not very good at large schemas with more than 10s of thousands of tables but we've got projects un
79.
by vvern 5y ago
cockroachdb implements subtransactions and I don't think any of these issues apply to cockroach's implementation.
80.
by vvern 5y ago
It feels to me like this just a kludge added to deal with a lack of a stateful iterator on top of the tree. In this use case, the author indicates that it is about exploiting locality in a request path. Imagine that the tree offered an iter
81.
by vvern 5y ago
> Additionally, out of band coordination, most likely via human beings , is required to make sure that all potential readers of the transactional writes are also transaction-aware. Subtle burn on the mongo and its client-coordinated ses
82.
by vvern 5y ago
> “If the GPS constellation fails, TrueTime fails (or gets really slow)". To deal with this threat model, spanner has atomic clocks in each data center, named “Armageddon time masters”, the most badass name of a component in any dis
83.
by vvern 5y ago
IIRC fdb is fully optimistic concurrency control. It doesn't do any locking. If you have workloads which are highly contended, you'll need to do something in the layer above to coordinate. Otherwise, performance will be unbearable
84.
by vvern 5y ago
This is an open source release of the database with the highest ever certified TPCC result. Horizontal scalability might be a bold claim but they’ve got some evidence behind it. tpc.org/tpcc/results/tpcc_perf_results5.asp?res
85.
by vvern 5y ago
> it's still near-impossible to get a usable dump of the in-memory object graph This, lately, is my #1 gripe. I just cannot get viewcore to provide valuable insights. Anybody know of any projects in this space?
86.
by vvern 5y ago
My intuition is that later versions of crdb are more like 1/3rd the efficiency of Postgres per core. GC is some of that but I don’t think it’s all that much. Everything has trade offs. Go is not the easiest language in which to write h
87.
by vvern 6y ago
This is timely. Cockroach has recently decided to lint against the go statement. We also don’t like errgroup because its context propagation is implicit. See https://github.com/cockroachdb/cockroach/pull/62243
88.
by vvern 6y ago
Undefined behavior ;)
89.
CockroachCloud Free Forever Tier(cockroachlabs.com)
5 points by vvern 6y ago|0 comments
90.
by vvern 6y ago
It's not the hardest thing to scale anymore. Engineering teams are pretty hard to scale.

More ›