yugabyte-db VS postgres-ha

It is possible to apply Batched Nested Loop but with additional code that checks the range of the outer bigint and compare it only if it matches the range of integer. This has been added in YugabyteDB 2.21 with #20715 YSQL: Allow BNL on joins over different integer types to help migrations from PostgreSQL with such datatype inconsistencies.

It can be done like https://github.com/yugabyte/yugabyte-db/ has.

Therefore, interleaving or co-partitioning is probably not necessary, and would reduce agility and scalability more than improving the performance. Unless you have a good reason for it that you can share on Issue #79. But, first, test and tune the queries to see if you need something else.

This is a workaround, not supported. I've opened the following issue to get it solve in the YugabyteDB deployment: https://github.com/yugabyte/yugabyte-db/issues/19389

Note that there may still be a need for bitmaps, especially with disjunctions (OR) as the following is about conjunction (AND), and it can still be implemented, differently than PostgreSQL. This is tracked by #4634.

FROM almalinux:8 as build RUN dnf -y update &&\ dnf groupinstall -y 'Development Tools' # get YugabyteDB sources ARG YB_TAG=2.18 RUN git clone --branch ${YB_TAG} https://github.com/yugabyte/yugabyte-db.git WORKDIR yugabyte-db # install dependencies and compilation tools RUN dnf install -y https://dl.fedoraproject.org/pub/epel/epel-release-latest-8.noarch.rpm RUN dnf -y install epel-release libatomic rsync python3-devel cmake3 java-1.8.0-openjdk maven npm golang gcc-toolset-12 gcc-toolset-12-libatomic-devel patchelf glibc-langpack-en ccache vim wget python3.11-devel python3.11-pip clang ncurses-devel readline-devel libsqlite3x-devel RUN mkdir /opt/yb-build RUN chown "$USER" /opt/yb-build # Install Python 3 RUN alternatives --remove-all python3 RUN alternatives --remove-all python RUN alternatives --install /usr/bin/python python /usr/bin/python3.11 3 RUN alternatives --install /usr/bin/python3 python3 /usr/bin/python3.11 3 # add #include "pg_yb_utils.h" to src/postgres/src/pl/plpython/plpy_procedure.c RUN sed -e '/#include "postgres.h"/a#include "pg_yb_utils.h"' -i src/postgres/src/pl/plpython/plpy_procedure.c # if using python > 3.9 remove #include and #include from src/postgres/src/pl/plpython/plpython.h RUN sed -e '/#include /d' -e '/#include /d' -i src/postgres/src/pl/plpython/plpython.h # add '--with-python', to python/yugabyte/build_postgres.py under the configure_postgres method RUN sed -e "/'\.\/configure',/a\ '--with-python'," -i python/yugabyte/build_postgres.py # Build and package the release RUN YB_CCACHE_DIR="$HOME/.cache/yb_ccache" ./yb_build.sh -j$(nproc) --clean-all --build-yugabyted-ui --no-linuxbrew --clang15 -f release RUN chmod +x bin/get_clients.sh bin/parse_contention.py bin/yb-check-consistency.py RUN YB_USE_LINUXBREW=0 ./yb_release --force WORKDIR / RUN mv /yugabyte-db/build/yugabyte*.tar.gz /yugabyte.tgz

You have seen me using the official YugabyteDB Docker image extensively. This image is suitable for various purposes, including labs, development, testing, and even production. In the past, we used to create it internally due to its seamless integration with our build process. However, some companies prefer to construct the image on their own, which is indeed a commendable practice. After all, it's not advisable to run random images with root privileges on your servers. As a result, we have made a significant alteration by introducing a refined Dockerfile to our Github repository.

Of course, I can guess from the function names, but YugabyteDB is Open Source and I can search for them. What happens here is that I didn't declare a Primary Key for my table and then an internal one (ybctid) is generated, because secondary indexes need a key to address the table row. This ID generation calls /dev/urandom. I made this simple example to show that low-level traces can give a clue about high level data model problems.

To show that it is not difficut to understand what you run, when in a PostgreSQL-compatible database, I'll look at the HammerDB benchmark connected to YugabyteDB. HammerDB has no specific code for it but YugabyteDB is PostgreSQL-compatible (it uses PostgreSQL code on top of distributed storage and transaction).

Fly Postgres is just a Fly.io app. You can see the source code for it right here:

https://github.com/fly-apps/postgres-ha

It has some direct `flyctl` integration (which is also open source), but it's not doing anything you can't do yourself if you want.

We do not! You have full administrative access to your postgres. You can create offsite replicas, or even fork the Postgres app we use and deploy over your Fly.io installed postgres: https://github.com/fly-apps/postgres-ha

RDS preventing external streaming replicas is the most annoying thing ever.

Our Postgres is not an RDS replacement. Lots of devs use RDS with Fly. In fact, Postgres on Fly is just a normal Fly app: https://github.com/fly-apps/postgres-ha

Ultimately, we think devs are better off if managed database services come from companies who specialize in those DBs. First party managed DBs trend towards mediocre, all the interesting Postgres features come from Heroku/Crunchy/Timescale/Supabase.

So we're "saving" managed Postgres for one of those folks. For the most part, they're more interested in giving AWS money because very large potential customers do. At some point, though, we'll be big enough to be attractive to DB companies.

Our postgres clusters are just a Fly app: https://github.com/fly-apps/postgres-ha

You could run your own PG by modifying that app. Right now we're calling it "automated" and not managed, though. All alerts about health and other issues go straight to customers, we don't have DBAs that will touch these things yet.