---
tags: rough-notes
---
# Environments & Communities

graph oais {
    rankdir=BT;
    
    node [color=Black,fontname=Arial,shape="box",style="rounded,filled",fillcolor="grey:white", gradientangle=90,width="1.5",height="0.5"];
    
    man [label="Management"];
    
    nodesep=1;
    sip [label="Producer"];
    aip [label="OAIS\n(archive)", height="1.0"];
    dip [label="Consumer"];
    
    sip -- aip -- dip;
    man -- aip;
    
    {rank=same; aip, dip, sip}
}

I think this (below) is more 'architectures', as the point is that those outside don't care how what's going in inside. i.e. the 'inner SIP' is not a SIP etc.

digraph oais_detail {
    rankdir=BT;
    
    node [color=Black,fontname=Arial,shape=box,style=filled,fillcolor="#eeeeee",height="0.4",width="0.5"];
    
    sip [label="SIP"];
    dip [label="DIP"];
    
    node [color=Black,fontname=Arial,shape="box",style="rounded,filled",fillcolor="grey:white", gradientangle=90,width="1.5",height="0.5"];
    
    man [label="Management"];
    
    pro [label="Producer"];
    arc [label="OAIS\n(archive)", height="1.0"];
    con [label="Consumer"];
    
    pro -> sip -> arc -> dip -> con;
    man -> arc;
    
    {rank=same; dip, sip, pro, arc, con}
}

## Preservation As A Service

digraph upstream_fork {
    rankdir=TB;
    newrank=true;
    
    node [color=Black,fontname=Arial,shape=box,style=filled,fillcolor="#eeeeee"];
    edge [];
    
    subgraph cluster1 {
      label="Publisher" fontname=Arial fillcolor="#bfffbf" style=filled color="#666666";    
      rank=same;
      src [label="Sources\n(≅SIP)"]
      up [label="Content\nManagment\n(≅AIP)"];
      dip [label="Publications\n(≅DIP)"];
      src -> up->dip;
    }
    
    subgraph cluster2 {
      label="Archive" fontname=Arial fillcolor="#ffffbf" style=filled color="#666666";
      rank=same;
      sip [label="SIP"];
      aip [label="AIP"];
      adip [label="DIP?", color="#777777", fontcolor="#777777", fillcolor="#f8f8f8", style=filled];
      color=black;

      sip -> aip;
      aip -> adip [color="#777777", style=dashed];
    }

    up:s->sip:n [constraint=true];
    
}

digraph upstream_fork {
    rankdir=TB;
    newrank=true;
    
    node [color=Black,fontname=Arial,shape=box,style=filled,fillcolor="#eeeeee"];
    edge [];
    
    subgraph cluster1 {
      label="Publisher" fontname=Arial fillcolor="#bfffbf" style=filled color="#666666";
      rank=same;
      src [label="Sources\n(≅SIP)"]
      up [label="Content\nManagment\n(≅AIP)"];
      dip [label="Publications\n(≅DIP)"];
      src -> up->dip;
    }
    
    subgraph cluster2 {
      label="Archive" fontname=Arial fillcolor="#ffffbf" style=filled color="#666666";
      rank=same;
      sip [label="SIP"];
      aip [label="AIP"];
      adip [label="DIP?", color="#777777", fontcolor="#777777", fillcolor="#f8f8f8"];
      color=black;

      sip -> aip;
      aip -> adip [color="#777777", style=dashed];
    }

    dip:s->sip:n [constraint=true];
    
}

Something else

#### Normalization on Ingest

Many organizations follow a strategy based on _Preservation Formats_ -- file formats that are considered to be better for long-term sustainability. In other cases, it's more a case of diversification, but where (unlike the Internet Archive example above) all the derivation is done before ingest.

In one case, the decision had been taken to normalize documents to PDF/A. The originals would be kept, but the PDFs would be used as the 'access copies' now, and were also expected to act as preservation copies over the medium term. The originals were not made available.

digraph longtime_dip {
    rankdir=LR;
    
    node [color=Black,fontname=Arial,shape=box,style=filled,fillcolor="#eeeeee"];
    edge [];
    
      sip [label="SIP"];
      subgraph cluster1 {
        label="AIP" fontname=Arial fillcolor="#ffffbf" style=filled color="#666666";
        dip [label="Preservation Format"];
        aip [label=Original];
      }
      adip [label="DIP"];

      sip -> aip:w;
      sip -> dip:w;
      dip -> adip;
}

The same thing can happen with other kinds of content, turning everything into TIFF or JPEG2000 or MPEG-4 or WAV.

While this is notionally for preservation purposes, it's usually _also_ because reducing the range of formats helps limit the complexity of the _current_ access system.

So, while the DIP is being generated from the Preservation Copy part of the AIP, the link between the Original and the Preservation Copy is less certain. i.e. you now have to prove the Preservation Copy contains everything you need from the Original, rather than that being enforced by the system design.

## web archive fragment

In the case of playing back archived web pages, current implementations ensure most of the necessary modification happens in the user's browser. Where that approach is impractical transformations are handled as part of the access system, such as the generation and caching of screenshots of archived pages. Similarly, format conversion can happen in the browser or in the access system, as appropriate.