diff options
| -rw-r--r-- | content/design.tex | 45 | ||||
| -rwxr-xr-x | images/ref_architecture.pdf | bin | 698876 -> 653496 bytes | |||
| -rw-r--r-- | images/ref_architecture.png | bin | 0 -> 879603 bytes |
3 files changed, 42 insertions, 3 deletions
diff --git a/content/design.tex b/content/design.tex index a41d84d..a207089 100644 --- a/content/design.tex +++ b/content/design.tex @@ -50,16 +50,55 @@ Based on the identified stakeholders we list 6 potential use-cases for a predict Congestion management and traffic routing can effectively benefit from digital twinning. Detecting bottlenecks, adjusting datacenter protocols and calibrating switches and interconnects is already a important use-case for \gls{dcdt}~\cite{DBLP:conf/sigcomm/HongWDSSHZY21}. \item \textbf{Virtual Prototyping} \\ - + Digital twinning can be used to provide insight into the system before changes are made. + Using a \gls{dcdt}, the operators can change, model and shape the datacenter to estimate their effect. + Virtual prototyping encompasses interacting with an existing datacenter model, or with a proof-of-concept, not yet constructed warehouse. \item \textbf{Monitoring and Visualization} \\ + 3D visualizations and dashboards are of the utmost importance to all stakeholders, due to the insights they provide. + With real-time data ingestion and the two-way feedback loop, \gls{dcdt}s can empower descriptive analytics. + This use-case already shapes many existing systems (see \Cref{tab:dt_features_comparison}). \end{enumerate} \subsection{Functional Requirements}\label{sss:functional_requirements} +Based on a subset of the above use-cases, we formulate the functional and non-functional requirements for \mysystem: \begin{enumerate}[label=\textbf{FR\arabic* --},align=left] - \item \textbf{Datacenter} \\ + \item \textbf{The system should be able to handle workloads of arbitrary size.} \\ + Existing systems range from Cloud through the Edge to HPC digital twins. + Therefore, \mysystem must support workloads similar in length and type to the commercial setting. + Without \textbf{FR1}, \mysystem will be incomplete, and like the majority of the \Cref{tab:dt_features_comparison} systems, its use-case will be niche. + \textbf{FR1} is necessary to avoid overly-specializing the \gls{dcdt}. + \item \textbf{The system should support failure detection.}\\ + Failures are of crucial concern to datacenter operators. + The system detect and report failures to datacenter operators. + Without \textbf{FR2}, the system will not be able to help datacenter operators meet \gls{sla}s. + Including \textbf{FR2} is necessary to ensure that failures, which which can cause financial penalties, are detected in a timely manner so as to meet the different \gls{sla}s. + \item \textbf{The system should be capable of long-term and short-term data storage.}\\ + \textbf{FR3} is necessary for \textbf{FR4} and \textbf{FR5}. + Without \textbf{FR3}, the system cannot support \gls{oda} techniques. + A system cannot be considered a \gls{dt} without insights stemming from accurate data analytics~\cite{DBLP:usdoe/report/AP26894}. + \gls{nasem} digital twin definition requires both real-time insights, and guidance based on historical-patterns. + Therefore, it is imperative to ensure \textbf{FR3}. + \item \textbf{The system should support real-time descriptive data analytics.}\\ + There are many types of data analytics present is existing deployments~\cite{DBLP:conf/wosp/SumanCNTMI24}. + In order to scope down the project, we only select several use-cases for \gls{my_system}. + Out of the 4 prime analytics type, we find in the literature survey the predictive analytics to be the most urgently needed, and descriptive analytics to be the most prevailing type of data processing. + Therefore, in order to achieve performance on par with previous systems, and to ensure our system meets the official \gls{nasem} definition, \textbf{FR4} is needed. + \item \textbf{The system should enable predictive data analytics.}\\ + The system should help future researchers incorporate predictive data analytics engines, regardless of the statistical modelling technique. + This is our novel contribution to the scientific field of \gls{dcdt}s. + Without \textbf{FR5}, we miss the aim of our entire project. + \item \textbf{The system should be able to process arbitrary amounts of telemetry.}\\ + The size of the \gls{ai} economy is expected to grow~\cite{DBLP:journals/computer/AthavaleBBMMPS24}. + As a result, both current and future datacenters will generate huge amount of data. + Our system must be capable of ingesting and processing different metrics regardless of the volume of incoming messages. + Without \textbf{FR6}, we hinder the adoption of \mysystem in modern and future datacenters. + \item \textbf{The model should be capable of clear data visualization.} + The system must have a user-friendly, visual interface for data analytics, and support them in real-time. + Without \textbf{FR7}, we exclude important stakeholders such as customers and users from the potential benefactors of our system. \end{enumerate} \subsection{Non-functional Requirements}\label{sss:non_functional_requirements} +In addition to the functional requirements, we also present non-functional requirements for \mysystem: \begin{enumerate}[label=\textbf{NFR\arabic* --},align=left] \item \textbf{Datacenter} \\ \end{enumerate} @@ -67,7 +106,7 @@ Based on the identified stakeholders we list 6 potential use-cases for a predict \section{Design of \emph{Sunfish}}\label{ss:design_of_mysystem} \begin{figure}[t] - \includegraphics[width=\linewidth]{images/ref_architecture.pdf} + \includegraphics[width=\linewidth]{images/ref_architecture.png} \caption{ The predictive datacenter digital twin reference architecture. We call the system \emph{Sunfish}. The architecture was designed with the \emph{AtLarge Design Process}~\cite{DBLP:conf/icdcs/IosupVTETBFMT19} over several iterations in the past months} diff --git a/images/ref_architecture.pdf b/images/ref_architecture.pdf Binary files differindex eb46624..670a813 100755 --- a/images/ref_architecture.pdf +++ b/images/ref_architecture.pdf diff --git a/images/ref_architecture.png b/images/ref_architecture.png Binary files differnew file mode 100644 index 0000000..d791545 --- /dev/null +++ b/images/ref_architecture.png |
