README.rst 4.02 KB
Newer Older
Kirill Smelkov's avatar
Kirill Smelkov committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
==========================================
 Wendelin.core - Out-of-core NumPy arrays
==========================================

Wendelin.core allows you to work with arrays bigger than RAM and local disk.
Bigarrays are persisted to storage, and can be changed in transactional manner.

In other words bigarrays are something like `numpy.memmap`_ for numpy.ndarray
and OS files, but support transactions and files bigger than disk. The whole
bigarray cannot generally be used as a drop-in replacement for numpy arrays,
but bigarray *slices* are real ndarrays and can be used everywhere ndarray can
be used, including in C/Cython/Fortran code. Slice size is limited by
virtual address-space size, which is ~ max 127TB on Linux/amd64.

The main class to work with is `ZBigArray` and is used like `ndarray` from
`NumPy`_:

1. create array::

    from wendelin.bigarray.array_zodb import ZBigArray
    import transaction

    # root is connected to opened database
    root['A'] = A = ZBigArray(shape=..., dtype=...)
    transaction.commit()

2. view array as a real ndarray::

    a = A[:]        # view which covers all array, if it fits into address-space
    b = A[10:100]

   data for views will be loaded lazily on memory access.

3. work with views, including using C/Cython/Fortran functions from NumPy
   and other libraries to read/modify data::

    a[2] = 1
    a[10:20] = numpy.arange(10)
    numpy.mean(a)

   | the amount of modifications in one transaction should be less than available RAM.
   | the amount of data read is limited only by virtual address-space size.

4. data can be appended to array in O(δ) time::

    values                  # ndarray to append of shape  (δ,)
    A.append(values)

   and array itself can be resized in O(1) time::

    A.resize(newshape)

5. changes to array data can be either discarded or saved back to DB::

    transaction.abort()     # discard all made changes
    transaction.commit()    # atomically save all changes



When using NEO_ or ZEO_ as a database, bigarrays can be simultaneously used by
several nodes in a cluster.


Please see `demo/demo_zbigarray.py`__ for a complete example.

66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109
__ demo/demo_zbigarray.py


Current state and Roadmap
=========================

Wendelin.core works in real life for workloads Nexedi_ is using in production,
including 24/7 projects. We are, however, aware of the following
limitations and things that need to be improved:

- wendelin.core is currently not very fast
- there are big - proportional to input in size - temporary array allocations
  in third-party libraries (NumPy_, `scikit-learn`_, ...) which might practically
  prevent processing out-of-core arrays depending on the functionality used.

Thus

- we are currently working on improved wendelin.core design and implementation,
  which will use kernel virtual memory manager (instead of one implemented__ in__
  userspace__) with arrays backend presented to kernel via FUSE as virtual
  filesystem implemented in Go.

__  https://lab.nexedi.com/nexedi/wendelin.core/blob/master/include/wendelin/bigfile/virtmem.h
__  https://lab.nexedi.com/nexedi/wendelin.core/blob/master/bigfile/virtmem.c
__  https://lab.nexedi.com/nexedi/wendelin.core/blob/master/bigfile/pagefault.c

In parallel we will also:

- try wendelin.core 1.0 on large data sets
- identify and incrementally fix big-temporaries allocation issues in NumPy and
  scikit-learn

We are open to community help with the above.


Additional materials
====================

- Wendelin.core tutorial__
- Slides__ (pdf__) from presentation about wendelin.core in PyData Paris 2015

__  https://www.nexedi.com/wendelin-Core.Tutorial.2016
__  http://www.wendelin.io/NXD-Wendelin.Core.Non.Secret/asEntireHTML
__  http://www.wendelin.io/NXD-Wendelin.Core.Non.Secret?format=pdf
110

Kirill Smelkov's avatar
Kirill Smelkov committed
111 112

.. _NumPy:          http://www.numpy.org/
113
.. _scikit-learn:   http://scikit-learn.org/
Kirill Smelkov's avatar
Kirill Smelkov committed
114 115 116
.. _numpy.memmap:   http://docs.scipy.org/doc/numpy/reference/generated/numpy.memmap.html
.. _NEO:            http://www.neoppod.org/
.. _ZEO:            https://pypi.python.org/pypi/ZEO
117
.. _Nexedi:         https://www.nexedi.com/