Add talk made on 2023-01-18 #482

Merged
lx merged 9 commits from talk-tocatta-2023-01-18 into main 2023-01-20 11:40:09 +00:00
2 changed files with 51 additions and 38 deletions
Showing only changes of commit d44e8366e7 - Show all commits

View file

@ -506,6 +506,39 @@
\hspace{1em}
\end{frame}
\begin{frame}
\frametitle{Consensus vs weak consistency}
\begin{center}
\textbf{From a theoretical point of view:}\\
\end{center}
\vspace{2em}
\hspace{1em}
\begin{minipage}{6.5cm}
\underline{Consensus-based systems:}
\vspace{1em}
Require \textbf{additionnal assumptions} such as a fault detector or a strong RNG\\
(FLP impossibility theorem)
\end{minipage}
\hfill
\begin{minipage}{6.5cm}
\underline{Weakly consistent systems:}
\vspace{1em}
Can be implemented in \textbf{any\\asynchronous message passing\\distributed system} with node crashes
\end{minipage}
\hspace{1em}
\vspace{3em}
\begin{center}
They represent \textbf{different classes of computational capability}\\
\end{center}
\end{frame}
\begin{frame}
\frametitle{Consensus vs weak consistency}
\begin{center}
@ -531,7 +564,7 @@
\vspace{1em}
\textbf{Limited objects such as CRDTs}\\(conflict-free replicated data types)
\textbf{Only CRDTs}\\(conflict-free replicated data types)
\vspace{1em}
Part of the complexity is \textbf{reported to the consumer of the API}\\~
@ -539,38 +572,6 @@
\hspace{1em}
\end{frame}
\begin{frame}
\frametitle{Consensus vs weak consistency}
\begin{center}
\textbf{From a theoretical point of view:}\\
\end{center}
\vspace{2em}
\hspace{1em}
\begin{minipage}{6.5cm}
\underline{Consensus-based systems:}
\vspace{1em}
Require \textbf{additionnal assumptions} such as a fault detector or a strong RNG\\~
\end{minipage}
\hfill
\begin{minipage}{6.5cm}
\underline{Weakly consistent systems:}
\vspace{1em}
Can be implemented in \textbf{any asynchronous message passing distributed system}
\end{minipage}
\hspace{1em}
\vspace{3em}
\begin{center}
They represent \textbf{different classes of computational capability}
\end{center}
\end{frame}
\begin{frame}
\frametitle{Understanding the power of consensus}
\textbf{Consensus:} an API with a single operation, $propose(x)$
@ -608,7 +609,7 @@
$\to$ the API cannot be implemented in a weakly consistent system
\vspace{2em}
\item \textbf{This API can be implemented using only weak primitives}\\
(e.g. a bunch of atomic registers)\\
(e.g. in the asynchronous message passing model with no further assumption)\\
$\to$ the API is strictly weaker than consensus\\
$\to$ we can implement it in Garage!
\end{itemize}
@ -625,7 +626,10 @@
\item \textbf{Performance issues:}
\vspace{1em}
\begin{itemize}
\item The leader is a \textbf{bottleneck} for all requests
\item Theoretical requirements (RNG, failure detector) translate into \textbf{practical costs}
\vspace{1em}
\item The leader is a \textbf{bottleneck} for all requests;\\
even in leaderless approaches, \textbf{all nodes must process all operations in order}
\vspace{1em}
\item Particularly \textbf{sensitive to higher latency} between nodes
\end{itemize}
@ -746,10 +750,19 @@
\end{frame}
\begin{frame}
\frametitle{The hard parts we don't address (yet!)}
\frametitle{A hard problem: layout changes}
\begin{itemize}
\item Maintain consistency changes when nodes assigned to a partition change:\\
\item TODO
\item We rely on quorums $k > n/2$ within each partition:\\
$$n=3,~~~~~~~k\ge 2$$
\item When rebalancing, the set of nodes responsible for a partition can change:\\
$$\{n_A, n_B, n_C\} \to \{n_A, n_D, n_E\}$$
\vspace{.01em}
\item During the rebalancing, $D$ and $E$ don't yet have the data,\\
~~~~~~~~~~~~~~~~~~~and $B$ and $C$ want to get rid of the data to free up space\\
\vspace{.2em}
$\to$ quorums only within the new set of nodes don't work\\
$\to$ how to coordinate? \textbf{currently, we don't...}
\end{itemize}
\end{frame}